A bed equipped with a sensor feature unit for determining the snoring and respiratory parameters of two sleepers.

Abstract

A mattress supports a left-side user and a right-side user. At least one acoustic sensor is configured to sense acoustic energy in the environment of the left-side user and the right-side user. At least one pressure sensor is configured to sense pressure applied by a left-side user to a left side and by a right-side user to a right side. A controller is configured to receive at least one acoustic stream from the at least one acoustic sensor and at least one pressure stream from the at least one pressure sensor. A left-side snoring / breathing parameter and a right-side snoring / breathing parameter are generated. The controller is further configured to send a command to a controllable device to operate the controllable device in response to determining that a home automation rule includes a condition involving at least one of the left-side snoring / breathing parameter and the right-side snoring / breathing parameter.

Description

Technical Field

[0001] This disclosure relates to the automation of consumer devices.

[0002] [Cross - Reference to Related Applications] This application claims the benefit of U.S. Patent Application No. 62 / 611,073, filed on December 28, 2017, and U.S. Patent Application No. 62 / 611,163, filed on December 28, 2017. The disclosures of the said U.S. patent applications are considered part of the disclosure of this application, and the entire contents thereof are incorporated herein by reference.

Background Art

[0003] Generally, a bed is furniture used as a place to sleep or relax. Many modern beds include a soft mattress on a bed frame. The mattress may include springs, foams and / or air chambers to support the weight of one or more occupants.

Summary of the Invention

[0004] In one feature, the bed system comprises a mattress for supporting a left-side user on the left side and a right-side user on the right side. The system further comprises at least one acoustic sensor configured to sense acoustic energy in the environment of the left-side and right-side users. The system further comprises at least one pressure sensor configured to sense pressure applied to the left side by the left-side user and to the right side by the right-side user. The system further comprises a controller configured to receive at least one acoustic stream from the at least one acoustic sensor. The at least one acoustic stream represents the acoustic energy sensed by the acoustic sensor. The controller further comprises at least one pressure stream from the at least one pressure sensor. The at least one pressure stream represents the pressure sensed by the pressure sensor. The controller further comprises generating left-side snoring / breathing parameters for the left-side user and right-side snoring / breathing parameters for the right-side user from the at least one acoustic stream and the at least one pressure stream. The controller is further configured to determine that a home automation rule includes a condition that includes at least one of the left-side snoring / breathing parameter and the right-side snoring / breathing parameter. The controller is further configured to send a command to a controllable device to drive the controllable device in response to the determination that the home automation rule includes a condition that includes at least one of the left-side snoring / breathing parameter and the right-side snoring / breathing parameter. Additional systems, methods, apparatus, and computer-readable media may be used.

[0005] The implementation may include, but may not include, any or all of the following features: The bed system may further include the controllable device, which may be configured to receive the command to drive the controllable device and, in response to the receipt of the command to drive the controllable device, to drive to change the user's environment. The at least one acoustic sensor may have two acoustic sensors. The at least one pressure sensor may have two pressure sensors. The at least one pressure sensor may have one pressure sensor configured to separately sense the pressure applied to the left side by a left-side user and separately sense the pressure applied to the right side by a right-side user. The at least one acoustic sensor may have a signal modifier and a digitizer, and the acoustic stream may be a digital data stream. The at least one pressure sensor may have a signal modifier and a digitizer, and the pressure stream may be a digital data stream. The controller may further be configured to combine one of the one or more acoustic streams with one of the one or more pressure streams to create a left-side combined stream, and to combine one of the one or more acoustic streams with one of the one or more pressure streams to create a right-side combined stream, and to generate left-side snoring / breathing parameters for a left-side user and right-side snoring / breathing parameters for a right-side user from the at least one acoustic stream and the at least one pressure stream using the left-side combined stream and the right-side combined stream. The controller may further be configured to generate left-side snoring / breathing parameters for a left-side user and right-side snoring / breathing parameters for a right-side user from the at least one acoustic stream and the at least one pressure stream using one sensing modality, and to verify the left-side snoring / breathing parameters and the right-side snoring / breathing parameters using a second sensing modality.The controller may further be configured to filter at least one of the at least one acoustic stream and the at least one pressure stream. The controller may further be configured to determine that at least one user is present on the mattress and that at least one user is asleep. The generation of the left-side snoring / breathing parameter and the right-side snoring / breathing parameter may be performed in response to the determination that at least one user is asleep. The controller may further be configured to use a machine learning classifier to generate a left-side snoring / breathing parameter for a left-side user and a right-side snoring / breathing parameter for a right-side user from the at least one acoustic stream and the at least one pressure stream.

[0006] The implementation may include, but may not include, any of the following features:

[0007] The technology for detecting breathing and snoring is improved. The disclosed technology can advantageously provide a stable monitoring solution for snoring and breathing sounds when two sleepers are sleeping in a bed (for example, a pair of adults sleeping on a king-size mattress in a house). The technology can advantageously provide the ability to monitor two sleepers simultaneously. Sleeper pressure sensor data indicating breathing behavior can be advantageously used in combination with acoustic sensor information. Automated snoring and breathing analysis can advantageously incorporate the user's presence and / or sleep state determined from the pressure data. The acoustics of snoring can be registered with the pressure signal to advantageously improve the reliability of acoustic-based snoring detection. Using both pressure and acoustic data can advantageously resist false alarms caused by ambient noise such as conversation, television, and street noise.

[0008] Other features, embodiments, and potential advantages will become apparent from the attached description and drawings. [Brief explanation of the drawing]

[0009] [Figure 1] Figure 1 shows an exemplary airbed system.

[0010] [Figure 2] Figure 2 is a block diagram of an example of various components of an airbed system.

[0011] [Figure 3] Figure 3 shows an exemplary environment that includes a bed communicating with multiple devices within and around the home.

[0012] [Figure 4A] Figures 4A and 4B are block diagrams of exemplary data processing systems that may be associated with a bed. [Figure 4B] Figures 4A and 4B are block diagrams of exemplary data processing systems that may be associated with a bed.

[0013] [Figure 5] Figures 5 and 6 are block diagrams of example motherboards that may be used in a data processing system that may be associated with a bed. [Figure 6] Figures 5 and 6 are block diagrams of example motherboards that may be used in a data processing system that may be associated with a bed.

[0014] [Figure 7] Figure 7 is a block diagram of an example of a daughterboard that may be used in a data processing system that may be associated with a bed.

[0015] [Figure 8] Figure 8 is a block diagram of an example of a motherboard without a daughterboard, which may be used in a data processing system that may be associated with a bed.

[0016] [Figure 9] Figure 9 is a block diagram of an example of a sensor array that may be used in a data processing system that may be associated with a bed.

[0017] [Figure 10] Figure 10 is a block diagram of an example of a controller array that can be used in a data processing system associated with a bed.

[0018] [Figure 11] Figure 11 is a block diagram of an example of a computing device that can be used in a data processing system associated with a bed.

[0019] [Figure 12] Figures 12 to 16 are block diagrams of an exemplary cloud service that can be used in a data processing system associated with a bed. [Figure 13] Figures 12 to 16 are block diagrams of an exemplary cloud service that can be used in a data processing system associated with a bed. [Figure 14] Figures 12 to 16 are block diagrams of an exemplary cloud service that can be used in a data processing system associated with a bed. [Figure 15] Figures 12 to 16 are block diagrams of an exemplary cloud service that can be used in a data processing system associated with a bed. [Figure 16] Figures 12 to 16 are block diagrams of an exemplary cloud service that can be used in a data processing system associated with a bed.

[0020] [Figure 17] Figure 17 is a block diagram of an example of automating peripheral devices around a bed using a data processing system associated with the bed.

[0021] [Figure 18] Figure 18 is a schematic diagram showing an example of a computing device and a mobile computing device.

[0022] [Figure 19A] Figures 19A and 19B are block diagrams of an exemplary system comprising two pressure sensors and two acoustic sensors, and block diagrams of data collected from such a system. [Figure 19B] Figures 19A and 19B are block diagrams of an exemplary system comprising two pressure sensors and two acoustic sensors, and block diagrams of data collected from such a system.

[0023] [Figure 20A] Figures 20A and 20B are block diagrams of an exemplary system comprising two pressure sensors and one acoustic sensor, and block diagrams of data collected from such a system. [Figure 20B] Figures 20A and 20B are block diagrams of an exemplary system comprising two pressure sensors and one acoustic sensor, and block diagrams of data collected from such a system.

[0024] [Figure 21] Figure 21 is a block diagram of an exemplary system with one pressure sensor and two acoustic sensors.

[0025] [Figure 22] Figure 22 is a flowchart of an exemplary process for operating a home automation device.

[0026] [Figure 23] Figure 23 is a pipeline diagram of an example of a pipeline (logic circuit chain) that may be used to remove non-snoring acoustic energy from an acoustic signal used to drive home automation.

[0027] [Figure 24A] Figures 24A and 24B are flowcharts illustrating an exemplary process for determining snoring / respiratory parameters. [Figure 24B]Figures 24A and 24B are flowcharts illustrating an exemplary process for determining snoring / respiratory parameters.

[0028] [Figure 25] Figure 25 is a swimlane diagram of an exemplary process for training and using a machine learning classifier to determine and classify snoring events in a bed.

[0029] [Figure 26] Figure 26 is a flowchart illustrating an exemplary process for training a classifier on pressure and / or acoustic signals.

[0030] [Figure 27] Figure 27 is a swimlane diagram of an exemplary process for training and using a machine learning classifier to determine and classify snoring events in a bed. [Modes for carrying out the invention]

[0031] In various drawings, similar reference symbols represent the same elements.

[0032] Acoustic and pressure sensing can be used in a bed where two people sleep to monitor and / or control a user's breathing and snoring at home. For example, large beds, such as those sold under names like "king" and "queen," are often used by two people. Acoustic and pressure sensors can be incorporated into the bed or the sleeping environment of the two users to simultaneously sense the snoring and breathing of two sleepers. Due to the physical proximity of the two users, a sensor positioned to detect acoustic phenomena created by one sleeper is likely to pick up acoustic phenomena created by the other sleeper. That is, the acoustic sensor on the left may pick up the snoring / breathing sounds created by the user on the right. Therefore, a sensor data analysis engine can be used to unravel the crosstalk created in the sensor data stream so that the sleep parameters of the two users can be accurately determined.

[0033] [Example Airbed Hardware]

[0034] Figure 1 shows an exemplary airbed system 100 including a bed 112. The bed 112 includes at least one air chamber 114, surrounded by an elastic boundary 116 and encapsulated by a durable cotton fabric 118 for the bed. The elastic boundary 116 may include any suitable material, such as foam.

[0035] As shown in Figure 1, the bed 112 may be a two-chamber design having first and second fluid chambers, such as a first air chamber 114A and a second air chamber 114B. In alternative embodiments, the bed 112 may include chambers for use with fluids other than air, as appropriate for the application. In some embodiments, such as a single bed or a children's bed, the bed 112 may include a single air chamber 114A or 114B, or multiple air chambers 114A and 114B. The first and second air chambers 114A and 114B may be in fluid communication with a pump 120. The pump 120 may electrically communicate with a remote control 122 via a control box 124. The control box 124 may include wired or wireless communication interfaces for communicating with one or more devices, including the remote control 122. The control box 124 may be configured to operate the pump 120 to increase or decrease the fluid pressure in the first and second air chambers 114A and 114B based on commands entered by the user using the remote control 122. In some implementations, the control box 124 is integrated into the housing of the pump 120.

[0036] The remote control 122 may include a display 126, an output selection mechanism 128, a pressure increase button 129, and a pressure decrease button 130. The output selection mechanism 128 may allow the user to switch the airflow generated by the pump 120 between the first and second air chambers 114A and 114B, thereby enabling control of multiple air chambers with a single remote control 122 and a single pump 120. For example, the output selection mechanism 128 can be a physical control unit (e.g., a switch or button) or an input control unit displayed on the display 126. Alternatively, separate remote control units may be provided for each air chamber, each including the ability to control multiple air chambers. The pressure increase button 129 and the pressure decrease button 130 may allow the user to increase or decrease the pressure in the air chamber selected by the output selection mechanism 128, respectively. Adjusting the pressure in the selected air chamber may result in a corresponding adjustment to the stiffness (hardness) of each air chamber. In some embodiments, the remote control 122 may be omitted or modified as appropriate depending on the application. For example, in some embodiments, the bed 112 may be controlled by a computer, tablet, smartphone, or other device that is wired or wirelessly connected to the bed 112.

[0037] Figure 2 is a block diagram of an example of various components of an airbed system. For example, these components may be used in an exemplary airbed system 100. As shown in Figure 2, the control box 124 may include a power supply 134, a processor 136, memory 137, a switching mechanism 138, and an analog-to-digital (A / D) converter 140. The switching mechanism 138 may be, for example, a relay or a solid-state switch. In some implementations, the switching mechanism 138 may be located in the pump 120 rather than in the control box 124.

[0038] Pump 120 and remote control 122 can communicate bidirectionally with control box 124. Pump 120 includes a motor 142, a pump manifold 143, a relief valve 144, a first control valve 145A, a second control valve 145B, and a pressure transducer 146. Pump 120 is fluidly connected to a first air chamber 114A and a second air chamber 114B via a first pipe 148A and a second pipe 148B, respectively. The first and second control valves 145A and 145B can be controlled by a switching mechanism 138 and are operable to adjust the fluid flow between pump 120 and the first and second air chambers 114A and 114B, respectively.

[0039] In some implementations, the pump 120 and the control box 124 may be supplied and packaged as a single unit. In some alternative implementations, the pump 120 and the control box 124 may be supplied as physically separate units. In some implementations, the control box 124, the pump 120, or both thereof, may be integrated into or contained within the bed frame or bed support structure supporting the bed 112. In some embodiments, the control box 124, the pump 120, or both thereof, may be located outside the bed frame or bed support structure (as shown in the example in Figure 1).

[0040] The exemplary airbed system 100 shown in Figure 2 includes two air chambers 114A and 114B and a single pump 120. However, other embodiments may include an airbed system having two or more air chambers and one or more pumps incorporated within the airbed system to control the air chambers. For example, a separate pump may be associated with each air chamber of the airbed system, or a single pump may be associated with multiple chambers of the airbed system. The separate pump may allow each air chamber to be inflated or deflated independently and simultaneously. Furthermore, additional pressure transducers may also be incorporated within the airbed system, for example, a separate pressure transducer may be associated with each air chamber.

[0041] During use, the processor 136 may, for example, send a depressurization command to one of the air chambers 114A, 114B, and the switching mechanism 138 may be used to convert the low-voltage command signal sent by the processor 136 into a higher operating voltage sufficient to activate the relief valve (safety valve) 144 of the pump 120 and open the control valves 145A, 145B. Opening the relief valve 144 may allow air to escape from the air chamber 114A or 114B through the respective air pipes 148A or 148B. During deflation, the pressure transducer 146 may transmit pressure readings to the processor 136 via the A / D converter 140. The A / D converter 140 may receive analog information from the pressure transducer 146 and convert this analog information into digital information usable by the processor 136. The processor 136 may transmit this digital signal to the remote control 122 to update the display 126 in order to inform the user of the pressure information.

[0042] As another example, the processor 136 may send a pressure increase command. The pump motor 142 may be energized in response to the pressure increase command and electronically actuate the corresponding valves 145A, 145B to supply air to the designated one of the air chambers 114A, 114B via the air pipes 148A, 148B. While air is being supplied to the designated air chamber 114A or 114B to increase the chamber's stiffness (rigidity), the pressure transducer 146 may sense the pressure in the pump manifold 143. In this case as well, the pressure transducer 146 may transmit the pressure reading to the processor 136 via the A / D converter 140. The processor 136 may use the information received from the A / D converter 140 to determine the difference between the actual pressure in the air chamber 114A or 114B and the desired pressure. The processor 136 may transmit the digital signal to the remote control 122 to update the display 126 in order to inform the user of the pressure information.

[0043] Generally speaking, during the expansion or contraction process, the pressure sensed within the pump manifold 143 can provide an approximation of the pressure in each air chamber that is in fluid communication with the pump manifold 143. An exemplary method for obtaining a pump manifold pressure reading substantially equal to the actual pressure in the air chambers comprises the steps of turning off the pump 120, allowing the pressures in the air chambers 114A or 114B and the pump manifold 143 to equalize, and then sensing the pressure in the pump manifold 143 using a pressure transducer 146. This allows sufficient time for the pressures in the pump manifold 143 and the chambers 114A or 114B to equalize, which can result in a pressure reading that is an accurate approximation of the actual pressure in the air chambers 114A or 114B. In some implementations, the pressures in the air chambers 114A and / or 114B may be continuously monitored using multiple pressure sensors (not shown).

[0044] In some implementations, the information collected by the pressure transducer 146 can be analyzed to determine various states of a person lying in bed 112. For example, the processor 136 may use the information collected by the pressure transducer 146 to determine the heart rate or respiratory rate of a person lying in bed 112. For example, the user may be lying on one side of bed 112, which includes a chamber 114A. The pressure transducer 146 may monitor pressure fluctuations in the chamber 114A, and this information may be used to determine the user's heart rate and / or respiratory rate. As another example, additional processing may be performed using the collected data to determine the person's sleep state (e.g., wakefulness, light sleep, deep sleep). For example, the processor 136 may determine when a person falls asleep, when they are asleep, and the person's various sleep states.

[0045] Additional information relating to the user of the airbed system 100, which can be determined using the information collected by the pressure transducer 146, includes the user's movements, the user's presence on the surface of the bed 112, the user's weight, the user's cardiac arrhythmia, and temporary apnea. Taking the detection of the user's presence as an example, the pressure transducer 146 may be used to detect the presence of a user on the bed 112, for example, by determining a change in total pressure and / or by one or more of the respiratory rate signal, heart rate signal, and / or other biometric signals. For example, a simple pressure sensing process may identify an increase in pressure as indicating the presence of a user on the bed 112. As another example, the processor 136 may determine that a user is present on the bed 112 if the detected pressure increases above a certain threshold (a threshold to indicate that a person or other object exceeding a certain weight is placed on the bed 112). As yet another example, the processor 136 may identify an increase in pressure in combination with detected slight rhythmic fluctuations in pressure as corresponding to the presence of a user on the bed 112. The presence of rhythmic fluctuations can be identified as being attributable to the user's breathing or heartbeat (or both). Breathing or heartbeat detection can distinguish the user present in bed from other objects (such as a suitcase) placed on the bed.

[0046] In some implementations, pressure fluctuations can be measured in the pump 120. For example, one or more pressure sensors may be placed in one or more internal cavities of the pump 120 to detect pressure fluctuations within the pump 120. Pressure fluctuations detected in the pump 120 may indicate pressure fluctuations in one or both chambers 114A and 114B. One or more sensors placed in the pump 120 may be in fluid communication with one or both chambers 114A and 114B, and the sensors may operate to determine the pressures within chambers 114A and 114B. The control box 124 may be configured to determine at least one vital sign (e.g., heart rate, respiratory rate) based on the pressure in chamber 114A or chamber 114B.

[0047] In some implementations, the control box 124 may analyze pressure signals detected by one or more pressure sensors to determine the heart rate, respiratory rate, and / or other vital signs of a user lying or sitting on the chamber 114A or chamber 114B. More specifically, when a user lies on the bed 112 positioned above the chamber 114A, each of the user's heart rate, respiration, and other movements may generate forces on the bed 112 that are transmitted to the chamber 114A. As a result of the force input to the chamber 114A due to the user's movements, waves may propagate through the chamber 114A into the pump 120. Pressure sensors located in the pump 120 may detect these waves, and the pressure signals output by the sensors may indicate heart rate, respiratory rate, or other information about the user.

[0048] Regarding sleep state, the airbed system 100 can determine the user's sleep state by using various biometric signals, such as heart rate, respiration, and / or user movement. While the user is sleeping, the processor 136 can receive one or more of the user's biometric signals (e.g., heart rate, respiration, and movement) and determine the user's current sleep state based on the received biometric signals. In some implementations, signals indicating pressure fluctuations in one or both chambers 114A and 114B may be amplified and / or filtered to allow for more accurate detection of heart rate and respiration.

[0049] The control box 124 may execute a pattern recognition algorithm or other calculation method based on the amplified and filtered pressure signal to determine the user's heart rate and respiratory rate. For example, the algorithm or calculation method may be based on the assumption that the heart rate portion of the signal has a frequency in the range of 0.5 to 4.0 Hz and the respiratory rate portion of the signal has a frequency in the range of less than 1 Hz. The control box 124 may also be configured to determine other user characteristics based on the received pressure signal, such as blood pressure, turning (rocking and rotating) movements, rolling movements, limb movements, weight, the presence or absence of the user, and / or the user's identity. A technique for monitoring a user's sleep using heart rate information, respiratory rate information, and other user information is disclosed in U.S. Patent Application Publication No. 2010 / 0170043 by Stephen J. Young et al., entitled “Apparatus for Monitoring Vital Signs.” The entire contents of that publication are incorporated herein by reference.

[0050] For example, a pressure transducer 146 may be used to monitor the air pressure in chambers 114A and 114B of bed 112. When the user on bed 112 is not moving, changes in the air pressure in air chamber 114A or 114B may be relatively small and may be due to breathing and / or heartbeat. However, when the user on bed 112 is moving, the air pressure in the mattress may fluctuate by a much larger amount. Therefore, the pressure signal generated by the pressure transducer 146 and received by the processor 136 may be filtered and shown as corresponding to movement, heartbeat, or breathing.

[0051] In some implementations, instead of performing data analysis within the control box 124 using processor 136, a digital signal processor (DSP) may be provided to analyze the data collected by the pressure transducer 146. Alternatively, the data collected by the pressure transducer 146 may be sent to a cloud-based computing system for remote analysis.

[0052] In some implementations, the exemplary airbed system 100 further comprises a temperature controller configured to raise, lower, or maintain the bed temperature for, for example, user comfort. For example, a pad may be placed on or as part of the top of the bed 112, or on or as part of the top of one or both of the chambers 114A and 114B. Air may be pushed through the pad to provide ventilation to cool the user of the bed. Conversely, the pad may include a heating element that can be used to keep the user warm. In some implementations, the temperature controller may receive temperature readings from the pad. In some implementations, separate pads may be used on different sides of the bed 112 (e.g., corresponding to the locations of the chambers 114A and 114B) to provide different temperature control on different sides of the bed.

[0053] In some implementations, a user of the airbed system 100 may use an input device such as a remote control 122 to input a desired temperature for the surface (or part of the surface) of the bed 112. The desired temperature may be encapsulated in a command data structure that includes the desired temperature and identifies the temperature controller as the desired controlled component. This command data structure may then be transmitted to the processor 136 via Bluetooth® or another suitable communication protocol. In various examples, the command data structure may be encrypted before transmission. The temperature controller may then be configured to increase or decrease the temperature of the pad in response to the temperature input by the user to the remote control 122.

[0054] In some implementations, data may be sent back from a component to the processor 136, or to one or more display devices such as the display 126. For example, the current temperature, bed pressure, current position of the base, or other information determined by the temperature controller's sensor element may be sent to the control box 124. The control box 124 may then send the received information to the remote control 122, where it can be displayed to the user (for example, on the display 126).

[0055] In some implementations, the exemplary airbed system 100 further comprises an adjustable base and an articulated controller configured to adjust the position of the bed (e.g., bed 112) by adjusting the adjustable base that supports the bed. For example, the articulated controller can adjust the bed 112 from a flat position to a position in which the head portion of the mattress of the bed is tilted upward (e.g., to facilitate the user sitting on the bed and / or watching television). In some implementations, the bed 112 includes several independently articulated sections. For example, portions of the bed corresponding to the positions of chambers 114A and 114B may be articulated independently of each other to allow one person to rest in a first position (e.g., a flat position) while a second person rests in a second position (e.g., a reclined position with the head tilted diagonally above the waist) while one person rests on the surface of the bed 112. In some implementations, two different beds (e.g., two twin beds placed next to each other) may have separate positions. The base of the bed 112 may include two or more zones that can be adjusted independently. The joint movement controller may also be configured to provide different levels of massage to one or more users on the bed 112.

[0056] [Example of a bed in a bedroom environment]

[0057] Figure 3 shows an exemplary environment 300 including a bed 302 that communicates with multiple devices in and around the home. In the illustrated example, the bed 302 includes a pump 304 for controlling the air pressure in two air chambers 306a and 306b (as previously described with respect to air chambers 114A-114B). The pump 304 further includes a circuit for controlling the inflation and deflation functions performed by the pump 304. The circuit is further programmed to detect fluctuations in the air pressure of air chambers 306a-b, and uses these detected fluctuations to identify the presence of user 308 in the bed, user 308's sleep state, user 308's movements, and user 308's biometric signals such as heart rate and respiratory rate. In the illustrated example, the pump 304 is located within the support structure of the bed 302, and a control circuit 334 for controlling the pump 304 is integrated with the pump 304. In some implementations, the control circuit 334 is physically separate from the pump 304 and communicates with the pump 304 wirelessly or via a wired connection. In some implementations, the pump 304 and / or the control circuit 334 are located outside the bed 302. In some implementations, various control functions can be performed by systems located in various physical locations. For example, the circuit for controlling the operation of the pump 304 may be located inside the pump casing of the pump 304, while the control circuit 334 for performing other functions related to the bed 302 may be located inside another part of the bed 302 or outside the bed 302. As another example, the control circuit 334 located inside the pump 304 may communicate with a remote control circuit 334 via a LAN or WAN (e.g., the Internet). As yet another example, the control circuit 334 may be included in the control box 124 shown in Figures 1 and 2.

[0058] In some implementations, one or more devices other than, or in addition to, pump 304 and control circuit 334 may be used to identify the user's presence, sleep state, movement, and biocharacteristic signals in the bed. For example, bed 302 may include a second pump in addition to pump 304, and each of the two pumps may be connected to one of the air chambers 306a and 306b, respectively. For example, pump 304 may be in fluid communication with air chamber 306b and control the expansion and contraction of air chamber 306b, and may detect user signals of the user located on air chamber 306b, such as presence, sleep state, movement, and biocharacteristic signals. On the other hand, the second pump may be in fluid communication with air chamber 306a and control the expansion and contraction of air chamber 306a, and may also detect user signals of the user located on air chamber 306a.

[0059] As another example, the bed 302 may include one or more pressure-sensitive pads or pressure-sensitive surface portions that are operable to detect movement, including the presence of a user, user movement, breathing, and heart rate. For example, a first pressure-sensitive pad may be incorporated into the surface of the bed 302 on the left side of the bed 302 where a first user normally sleeps, and a second pressure-sensitive pad may be incorporated into the surface of the bed 302 on the right side of the bed 302 where a second user normally sleeps. The movement detected by the one or more pressure-sensitive pads or pressure-sensitive surface portions may be used by the control circuit 334 to identify the user's sleep state, presence in the bed, or biometric signals.

[0060] In some implementations, information detected by the bed (e.g., motion information) is processed by a control circuit 334 (e.g., a control circuit 334 integrated with the pump 304) and provided to one or more user devices, such as a user device 310, for presentation to user 308 or other users. In the example shown in Figure 3, user device 310 is a tablet device. However, in some implementations, user device 310 may be a personal computer, a smartphone, a smart TV (e.g., TV 312), or another user device capable of wired or wireless communication with the control circuit 334. User device 310 may communicate with the control circuit 334 of bed 302 via a network or via direct point-to-point communication. For example, the control circuit 334 may be connected to a LAN (e.g., via a Wi-Fi router) and communicate with user device 310 via the LAN. In another example, both the control circuit 334 and user device 310 may be connected to the Internet and communicate via the Internet. For example, the control circuit 334 may connect to the internet via a WiFi router, and the user device 310 may connect to the internet via communication with a cellular communication system. As another example, the control circuit 334 may communicate directly with the user device 310 via a wireless communication protocol such as Bluetooth®. As yet another example, the control circuit 334 may communicate with the user device 310 via a wireless communication protocol such as ZigBee, Z-Wave, infrared, or other wireless communication protocols appropriate for the application. As yet another example, the control circuit 334 may communicate with the user device 310 via a wired connection such as a USB connector, serial / RS232, or other wired connection appropriate for the application.

[0061] The user device 310 may display various information and statistics related to sleep or user 308's interaction with bed 302. For example, the user interface displayed by the user device 310 may present information including the amount of sleep user 308 has over a certain period (e.g., one night, one week, one month), the amount of deep sleep, the ratio of deep sleep to restless sleep, the time elapsed between user 308 getting into bed and falling asleep, the total time spent in bed 302 over a given period, user 308's heart rate over a certain period, user 308's respiratory rate over a certain period, or other information related to user interaction with bed 302 by user 308 or one or more other users of bed 302. In some implementations, information from multiple users may be presented to the user device 310; for example, information from a first user located on air chamber 306a may be presented together with information from a second user located on air chamber 306b. In some implementations, the information presented on the user device 310 may change depending on the user's age. For example, the information presented on the user device 310 may evolve with the user's age, and different information may be presented on the user device 310 as the user 308 ages from child to adult.

[0062] The user device 310 may also be used as an interface for the control circuit 334 of the bed 302 to allow user 302 to input information. Information input by user 308 may be used by the control circuit 334 to provide better information to the user or to various control signals for controlling the functions of the bed 302 or other devices. For example, user 308 may input information such as weight, height, and age, and the control circuit 334 may use this information to provide the user with a comparison of the user's tracked sleep information with the sleep information of other people with similar weight, height, and / or age. As another example, user 308 may use the user device 310 as an interface to control the air pressure of the air chambers 306a and 306b, to control various reclining or tilting positions of the bed 302, to control the temperature of one or more surface temperature control devices of the bed 302, or to allow the control circuit 334 to generate control signals for other devices (as described in more detail below).

[0063] In some implementations, the control circuit 334 of the bed 302 (for example, the control circuit 334 integrated into the pump 304) may communicate with other first, second, or third-party devices or systems in addition to, or instead of, the user device 310. For example, the control circuit 334 may communicate with the television 312, the lighting system 314, the thermostat 316, the security system 318, or other household appliances such as the oven 322, the coffee maker 324, the lamp 326, and the night light 328. Other examples of devices and / or systems that the control circuit 334 may communicate with include a system for controlling the blinds 330, one or more devices for detecting or controlling the state of one or more doors 332 (e.g., detecting whether a door is open or not, detecting whether a door is locked or not, or automatically locking a door), and a system for controlling the garage door 320 (e.g., a control circuit 334 integrated with a garage door opener to identify the open / closed state of the garage door 320 and cause the garage door opener to open or close the garage door 320). Communication between the control circuit 334 of the bed 302 and other devices may occur via a network (e.g., LAN or the Internet) or as point-to-point communication (e.g., Bluetooth®, wireless communication, or wired connection). In some implementations, the control circuit 334 of different beds 302 may communicate with different sets of devices. For example, a kids' bed may not communicate with and / or control the same devices as an adult bed. In some embodiments, the bed 302 may evolve with the user's age such that the control circuit 334 of the bed 302 communicates with different devices as a function of the user's age.

[0064] The control circuit 334 may receive information and inputs from other devices / systems and use such received information and inputs to control the operation of the bed 302 or other devices. For example, the control circuit 334 may receive information from a thermostat 316 indicating the current ambient temperature of the house or room in which the bed 302 is located. The control circuit 334 may use this received information (along with other information) to determine whether to raise or lower the temperature of all or part of the surface of the bed 302. The control circuit 334 may then instruct the heating or cooling mechanism of the bed 302 to raise or lower the surface temperature of the bed 302. For example, user 308 may indicate a desired sleep temperature of 74 degrees Fahrenheit, while a second user of the bed 302 may indicate a desired sleep temperature of 72 degrees Fahrenheit. The thermostat 316 may indicate to the control circuit 334 that the current temperature of the bedroom is 72 degrees Fahrenheit. The control circuit 334 may recognize that user 308 has indicated a desired sleep temperature of 74 degrees Fahrenheit and may send a control signal to a heating pad on the user 308's side of the bed 302 to raise the temperature of a portion of the bed surface. It is positioned to raise the temperature of the user 308's sleeping surface to the desired temperature.

[0065] The control circuit 334 may also generate control signals to control other devices and propagate such control signals to those other devices. In some implementations, the control signals are generated based on information collected by the control circuit 334, including information about user interactions with the bed 302 by user 308 and / or one or more other users. In some implementations, information collected from one or more other devices other than the bed 302 is used when generating control signals. For example, information about environmental occurrences (e.g., ambient temperature, ambient noise level, ambient light level), time, year, day of the week, or other information may be used when generating control signals for various devices that communicate with the control circuit 334 of the bed 302. For example, information about the time may be combined with information about user 308's movement and presence in the bed to generate a control signal for the lighting system 314. In some implementations, instead of providing control signals to one or more other devices, or in addition to doing so, the control circuit 334 may transmit collected information (e.g., information related to user movement, presence in bed, sleep state, or user 308's bio-characteristic signals) to one or more other devices, allowing those devices to utilize the collected information when generating control signals. For example, the control circuit 334 of bed 302 may provide a central controller (not shown) with information regarding user interaction with bed 302 by user 308. The central controller may utilize the provided information to generate control signals for various devices, including bed 302.

[0066] Continuing with Figure 3, the control circuit 334 of bed 302 may generate control signals to control the operation of other devices in response to information collected by the control circuit 334, including the presence of user 308 in bed, the user's sleep state 308, and other factors, and may transmit such control signals to the other devices. For example, the control circuit 334 integrated with the pump 304 may detect features of the bed 302 mattress, such as an increase in pressure in the air chamber 306b, and use this detected increase in air pressure to determine that user 308 is on bed 302. In some implementations, the control circuit 334 may identify user 308's heart rate or respiratory rate to identify that the increase in pressure is due to a person sitting, lying down, or resting on bed 302, rather than an inanimate object (such as a suitcase) being placed on the bed. In some implementations, information indicating the user's presence in bed may be combined with other information to identify user 308's current or future possible state. For example, a user's presence in bed detected at 11:00 AM may indicate that the user is sitting in bed (e.g., tying their shoelaces or reading a book) and is not yet planning to sleep. On the other hand, a user's presence in bed detected at 10:00 PM may indicate that user 308 is in bed and intends to sleep soon. As another example, if control circuit 334 detects that user 308 has left bed 302 at 6:30 AM (e.g., that user 308 has woken up for the day), and then detects user 308's presence in bed at 7:30 AM, control circuit 334 may use (understand) this information not as an indication that user 308 intends to stay in bed for an extended period, but rather as the newly detected presence of the user in bed is likely to be temporary (e.g., while user 308 is tying their shoelaces before going to work).

[0067] In some implementations, the control circuit 334 may use the collected information (including information related to user interaction with bed 302 by user 308, environmental information, time information, and input received from user) to identify user 308's usage patterns. For example, the control circuit 334 may use information collected over a period of time indicating user 308's presence in bed and sleep state to identify the user's sleep patterns. For example, based on information indicating the user's presence collected over a week and user 308's biometric signaling, the control circuit 334 may identify that user 308 generally goes to bed between 9:30 p.m. and 10:00 p.m., generally falls asleep between 10:00 p.m. and 11:00 p.m., and generally wakes up between 6:30 a.m. and 6:45 a.m. The control circuit 334 may use the user's identification patterns to better process and identify user interaction with bed 302 by user 308.

[0068] For example, given the bed presence, sleep, and wake patterns of user 308 in the above example, if it is detected that user 308 is in bed at 3:00 p.m., the control circuit 334 may determine that the user's presence in bed is merely temporary and use this determination to generate a different control signal than one that would be generated if the control circuit 334 determined that user 308 was in bed in the evening. In another example, if the control circuit 334 detects that user 308 has gotten out of bed at 3:00 a.m., the control circuit 334 may use user 308's identification pattern to determine that the user was only temporarily awake (for example, to use the toilet or get a glass of water) and not to have gotten up for the day. In contrast, if the control circuit 334 identifies that user 308 got out of bed 302 at 6:40 a.m., the control circuit 334 may determine that the user has woken up for the day and may generate a different set of control signals than would be generated if it were determined that user 308 had only temporarily gotten out of bed (as in the case of user 308 getting out of bed 302 at 3:00 a.m.). For other users 308, getting out of bed 302 at 3:00 a.m. may be a normal time to wake up, so the control circuit 334 may learn and respond accordingly.

[0069] As described above, the control circuit 334 of bed 302 can generate control signals for the control functions of various other devices. The control signals can be generated, at least in part, based on the detected interaction between user 308 and bed 302, and other information including time, date, and temperature. For example, the control circuit 334 can communicate with television 312, receive information from television 312, and generate control signals to control the functions of television 312. For example, the control circuit 334 can receive an indication from television 312 that television 312 is currently on. If television 312 is located in a different room from bed 302, the control circuit 334 can generate a control signal to turn off television 312 when it determines that user 308 has gone to sleep at night. For example, if the presence of user 308 on bed 302 is detected within a specific time range (e.g., between 8:00 PM and 7:00 AM) and lasts longer than a threshold time (e.g., 10 minutes), the control circuit 334 may use this information to determine that user 308 is in bed for sleep. If television 312 is on (indicated by communication received by control circuit 334 on bed 302 from television 312), control circuit 334 may generate a control signal to turn off television 312. The control signal can then be transmitted to television (e.g., via a directed communication link between television 312 and control circuit 334, or via a network). As another example, instead of turning off television 312 in response to the detection of a user's presence in bed, control circuit 334 may generate a control signal to lower the volume of television 312 by a predetermined amount.

[0070] As another example, when the control circuit 334 detects that user 308 has left bed 302 within a specified time range (for example, between 6:00 AM and 8:00 AM), it may generate a control signal to turn on television 312 and tune it to a pre-specified channel (for example, user 308 indicates a preference to watch the morning news when getting out of bed in the morning). The control circuit 334 may generate a control signal and transmit it to television 312 to turn on television 312 and tune it to a desired channel (which may be stored in the control circuit 334, television 312, or another location). As yet another example, when the control circuit 334 detects that user 308 has woken up that day, it may generate and transmit a control signal to turn on television 312 and start playback of a previously recorded program from a digital video recorder (DVR) communicating with television 312.

[0071] As another example, if the television 312 is in the same room as the bed 302, the control circuit 334 does not turn off the television 312 in response to detecting the user's presence in bed. Rather, the control circuit 334 may turn off the television 312 by generating and transmitting a control signal in response to determining that the user 308 is asleep. For example, the control circuit 334 may monitor the user 308's biometric signals (e.g., movement, heart rate, respiratory rate) to determine that the user 308 has fallen asleep. When it detects that the user 308 is asleep, the control circuit 334 generates and transmits a control signal to turn off the television 312. As another example, the control circuit 334 may generate a control signal to turn off the television 312 after a threshold time has elapsed since the user 308 fell asleep (e.g., 10 minutes after the user fell asleep). As yet another example, the control circuit 334 generates a control signal to lower the volume of the television 312 after determining that the user 308 is asleep. As yet another example, the control circuit 334, in response to the determination that user 308 is asleep, generates and transmits a control signal to gradually lower the volume of the television over a period of time, and then turns off the television.

[0072] In some implementations, the control circuit 334 can interact with other media devices such as computers, tablets, smartphones, and stereo systems. For example, when the control circuit 334 detects that user 308 is asleep, it may generate a control signal and send it to the user device 310, which may turn off the user device 310 or lower the volume of a video or audio file being played on the user device 310.

[0073] The control circuit 334 can further communicate with the lighting system 314, receive information from the lighting system 314, and generate control signals to control the functions of the lighting system 314. For example, when it detects the presence of a user on bed 302 for a longer period than a threshold time (e.g., 10 minutes) within a specific time frame (e.g., between 8:00 p.m. and 7:00 a.m.), the control circuit 334 of bed 302 may determine that user 308 is in bed for sleep. In response to this determination, the control circuit 334 may generate a control signal to turn off the lights in one or more rooms other than the room where bed 302 is located. The control signal can then be transmitted to the lighting system 314, which may execute it to turn off the lights in the indicated room. For example, the control circuit 334 may generate and transmit a control signal to turn off the lights in all general rooms, rather than other bedrooms. As another example, in response to the determination that user 308 is in bed for sleep, the control signal generated by the control circuit 334 may indicate that the lights in all rooms except the room where bed 302 is located should be turned off, and that one or more lights located outside the house, including bed 302, should also be turned off. Furthermore, in response to the determination that user 308 is in bed or that user 308 is asleep, the control circuit 334 may generate and transmit a control signal to turn on the night light 328. As yet another example, the control circuit 334 may generate a first control signal to turn off a first set of lights (e.g., lights in general rooms) in response to the detection of the user's presence in bed, and a second control signal to turn off a second set of lights (e.g., lights in the room where bed 302 is located) in response to the detection that user 308 is asleep.

[0074] In some implementations, in response to the determination that user 308 is in bed for sleep, the control circuit 334 of bed 302 may generate a control signal that causes the lighting system 314 to implement a sunset lighting pattern in the room where bed 302 is located. The sunset lighting pattern may include dimming the lights (gradually over time or abruptly) in combination with changing the color of the lighting in the bedroom environment, such as adding an amber hue to the bedroom lighting. The sunset lighting pattern may help user 308 fall asleep when the control circuit 334 determines that user 308 is in bed for sleep.

[0075] The control circuit 334 may also be configured to implement a sunrise illumination pattern when user 308 wakes up in the morning. The control circuit 334 may determine that user 308 has woken up for the day by detecting, for example, that user 308 has left bed 302 (i.e., is no longer in bed 302) during a specified time frame (e.g., between 6:00 a.m. and 8:00 a.m.). Alternatively, the control circuit 334 may monitor user 308's movement, heart rate, respiratory rate, or other biometric signals to determine that user 308 is awake even if user 308 has not left bed. If the control circuit 334 detects that the user is awake during the specified time frame, the control circuit 334 may determine that user 308 has woken up for the day. The specified time frame may be based, for example, on previously recorded information about the user's presence in bed collected over a period of time (e.g., two weeks). This may indicate that user 308 typically wakes up between 6:30 a.m. and 7:30 a.m. In response to the control circuit 334 determining that user 308 is awake, the control circuit 334 may generate a control signal to cause the lighting system 314 to implement a sunrise lighting mode in the bedroom where the bed 302 is located. The sunrise lighting mode may include, for example, turning on a light (e.g., lamp 326, or other lights in the bedroom). The sunrise lighting mode may further include gradually increasing the level of lighting in the room where the bed 302 is located (or one or more other rooms). The sunrise lighting mode may also include turning on only lights of a specified color. For example, the sunrise lighting mode may include illuminating the bedroom with blue light to gently assist user 308 in waking up and becoming active.

[0076] In some implementations, the control circuit 334 may generate different control signals to control the operation of one or more components, such as the lighting system 314, depending on the time when user interaction with the bed 302 is detected. For example, the control circuit 334 may use historical user interaction information about the interaction between user 308 and bed 302 to determine that user 308 usually falls asleep between 10:00 p.m. and 11:00 p.m. and wakes up between 6:30 a.m. and 7:30 a.m. Using this information, the control circuit 334 may generate a first set of control signals to control the lighting system 314 if it is detected that user 308 got out of bed at 3:00 a.m., and a second set of control signals to control the lighting system 314 if it is detected that user 308 got out of bed after 6:30 a.m. For example, if user 308 gets out of bed before 6:30 a.m., the control circuit 334 may turn on the lights that guide user 308 to the toilet. As another example, if user 308 gets out of bed before 6:30 a.m., the control circuit 334 may turn on lights to guide user 308 to the kitchen (this may include, for example, turning on the night light 328, the under-bed light, or the lamp 326).

[0077] As another example, if user 308 gets out of bed after 6:30 a.m., control circuit 334 may generate a control signal to cause the lighting system 314 to start sunrise lighting mode, or to turn on one or more lights in the bedroom or other rooms. In some implementations, if it is detected that user 308 gets out of bed before a designated morning wake-up time for that user, control circuit 334 may cause the lighting system 314 to turn on a light that is weaker (dimmer) than the light that would be turned on by the lighting system 314 if it were detected that user 308 gets out of bed after the designated morning wake-up time. Turning on only weak (dim) lighting when user 308 gets out of bed at night (i.e., before user 308's normal wake-up time) can prevent other residents of the house from being woken by the lights, while still allowing user 308 to see (provide a view) to reach the toilet, kitchen, or another destination within the house.

[0078] The history of user interaction information regarding the interaction between user 308 and bed 302 can be used to identify the user's sleep and wake timeframes. For example, the user's time spent in bed and sleep time can be determined for a set period (e.g., two weeks, one month). The control circuit 334 can then identify the typical time range or timeframe in which user 308 goes to bed, the typical timeframe in which user 308 falls asleep, and the typical timeframe in which user 308 wakes up (in some cases, the timeframe in which user 308 wakes up may differ from the timeframe in which user 308 actually gets out of bed). In some implementations, buffer time may be added to these timeframes. For example, if it is identified that the user typically goes to bed between 10:00 p.m. and 10:30 p.m., a 30-minute buffer may be added to the timeframe in each direction, and detection of the user getting into bed between 9:30 p.m. and 11:00 p.m. may be interpreted as user 308 going to bed at night. As another example, the detection of user 308's presence in bed within a time frame that begins 30 minutes before the earliest typical time user 308 goes to bed and extends until the user's typical wake-up time (e.g., 6:30 a.m.) may be interpreted as user 308 going to bed at night. For example, if a user typically goes to bed between 10:00 p.m. and 10:30 p.m., then if the user's presence in bed is detected at 12:30 a.m. (0:30 a.m.) on a given night, this may be interpreted as user 308 going to bed at night, even though it is beyond the user's typical bedtime frame, because it occurs before the user's usual wake-up time. In some implementations, different time frames are identified for different times of the year (e.g., bedtime is earlier in winter than in summer) or different days of the week (e.g., users wake up earlier on weekdays than on weekends).

[0079] The control circuit 334 can distinguish between short-term presence on the bed 302 (such as a nap) and prolonged sleep (such as at night) by sensing the duration of the user 308's presence. In some cases, the control circuit 334 can distinguish between short-term sleep (such as a nap) and prolonged sleep (such as at night) by sensing the duration of the user 308's sleep. For example, the control circuit 334 may set a time threshold so that if the user 308 is detected on the bed 302 for longer than the threshold, the user 308 is considered to have slept at night. In some cases, the threshold may be approximately 2 hours, so if the user 308 is detected on the bed 302 for more than 2 hours, the control circuit 334 registers it as a prolonged sleep event. In other cases, the threshold may be longer or shorter than 2 hours.

[0080] The control circuit 334 can detect recurring long-duration sleep events in order to automatically determine the user 308's typical bedtime range without requiring the user 308 to input a bedtime range. This allows the control circuit 334 to accurately estimate the times when the user 308 is likely to go to bed due to long-duration sleep events, regardless of whether the user 308 typically goes to bed using a traditional or non-traditional sleep schedule. The control circuit 334 can then use its knowledge of the user 308's bedtime range to control one or more components (including the bed 302 and / or non-bed peripherals) differently based on its perception of being in bed within or outside the bedtime range.

[0081] In some examples, the control circuit 334 may automatically determine the user 308's bedtime range without requiring user input. In some examples, the control circuit 334 may automatically determine the user 308's bedtime range in combination with user input. In some examples, the control circuit 334 may directly set the bedtime range according to user input. In some examples, the control circuit 334 may associate different bedtimes with different days of the week. In each of these examples, the control circuit 334 may control one or more components (such as the lighting system 314, thermostat 316, security system 318, oven 322, coffee maker 324, lamp 326, and night light 328) as a function of the detected presence in bed and the bedtime range.

[0082] The control circuit 334 can further communicate with the thermostat 316, receive information from the thermostat 316, and generate control signals to control the functions of the thermostat 316. For example, user 308 may indicate different temperature preferences at different times depending on user 308's sleep state or presence in bed. For example, user 308 may prefer ambient temperatures of 72 degrees Fahrenheit when getting out of bed, 70 degrees Fahrenheit when in bed but awake, and 68 degrees Fahrenheit when asleep. The control circuit 334 of bed 302 can detect user 308's presence in bed at night and determine that user 308 is asleep. In response to this determination, the control circuit 334 can generate a control signal to the thermostat to change the temperature to 70 degrees Fahrenheit. Next, the control circuit 334 can transmit this control signal to the thermostat 316. If the control circuit 334 detects that user 308 is asleep or in bed during the designated bedtime range, it may generate and transmit a control signal to cause the thermostat 316 to change the temperature to 68 degrees Fahrenheit. The following morning, if the control circuit 334 determines that the user has gotten up (for example, that user 308 got out of bed after 6:30 a.m.), it may generate and transmit a control signal to cause the thermostat 316 to change the temperature to 72 degrees Fahrenheit.

[0083] In some implementations, the control circuit 334 may also generate control signals to change the temperature of one or more heating or cooling elements on the surface of the bed 302 at various points in time, in response to user interaction with the bed 302 or at various pre-programmed times. For example, when the control circuit 334 detects that the user 308 has fallen asleep, it may activate a heating element to raise the temperature of one side of the bed 302's surface to 73 degrees Fahrenheit. As another example, when it determines that the user 308 has woken up for the day, the control circuit 334 may turn off the heating or cooling element. As yet another example, the user 308 may pre-program various times at which the bed surface temperature should rise or fall. For example, the user may program the bed 302 to raise the surface temperature to 76 degrees Fahrenheit at 10:00 p.m. and lower it to 68 degrees Fahrenheit at 11:30 p.m.

[0084] In some implementations, in response to the detection of user 308's presence in bed and / or the detection that user 308 is asleep, the control circuit 334 may cause the thermostat 316 to change the temperature of different rooms to different values. For example, in response to the determination that user 308 is in bed at night, the control circuit 334 may generate and transmit a control signal that causes the thermostat 316 to set the temperature of one or more bedrooms in the house to 72 degrees Fahrenheit and the temperature of other rooms to 67 degrees Fahrenheit.

[0085] The control circuit 334 may also receive temperature information from the thermostat 316 and use this temperature information to control the function of the bed 302 or other devices. For example, as described above, the control circuit 334 may adjust the temperature of the heating elements included in the bed 302 in response to the temperature information received from the thermostat 316.

[0086] In some implementations, the control circuit 334 may generate and transmit control signals to control other temperature control systems. For example, in response to a determination that user 308 has woken up for the day, the control circuit 334 may generate and transmit a control signal to activate the underfloor heating element. For instance, in response to a determination that user 308 has woken up for the day, the control circuit 334 may turn on the underfloor heating system in the master bedroom.

[0087] The control circuit 334 can further communicate with the security system 318, receive information from the security system 318, and generate control signals to control the functions of the security system 318. For example, in response to detection that user 308 has gone to sleep at night, the control circuit 334 can generate a control signal to cause the security system to activate or deactivate its security functions. The control circuit 334 can then transmit this control signal to the security system 318 to activate it. As another example, in response to determination that user 308 has woken up that day (for example, that user 308 is no longer in bed 302 after 6:00 a.m.), the control circuit 334 can generate and transmit a control signal to disable the security system 318. In some implementations, the control circuit 334 may, in response to the detection of the presence of user 308 in bed, generate and transmit a first set of control signals to the security system 318 to activate a first set of security functions, and in response to the detection that user 308 has fallen asleep, generate and transmit a second set of control signals to the security system 318 to activate a second set of security functions.

[0088] In some implementations, the control circuit 334 may receive alerts from the security system 318 (and / or a cloud service associated with the security system 318) and display these alerts to the user 308. For example, the control circuit 334 may detect that the user 308 is in bed at night and, in response, generate and transmit control signals to activate or deactivate the security system 318. The security system may then detect a security breach (e.g., someone opens the door 332 without entering a security code, or someone opens a window while the security system 318 is activated). The security system 318 may communicate the security breach to the control circuit 334 of the bed 302. In response to receiving the communication from the security system 318, the control circuit 334 may generate a control signal to warn the user 308 about the security breach. For example, the control circuit 334 may vibrate the bed 302. As another example, the control circuit 334 may articulate a part of the bed 302 (e.g., raise or lower the head section) to wake the user 308 and warn the user of a security breach. As yet another example, the control circuit 334 may generate and transmit a control signal to cause the lamp 326 to flash at regular intervals to warn the user 308 of a security breach. As yet another example, the control circuit 334 may warn the user 308 in one bed 302 of a security breach in another bed's bedroom, such as an open window in a child's bedroom. As yet another example, the control circuit 334 may send a warning to a garage door controller (e.g., to close and lock the door). As yet another example, the control circuit 334 may send a warning to deactivate the security.

[0089] The control circuit 334 can further generate and transmit control signals to control the garage door 320 and receive information indicating the state of the garage door 320 (i.e., whether it is open or closed). For example, in response to the determination that user 308 is in bed at night, the control circuit 334 can generate and transmit a request to a garage door opener or another device capable of sensing whether the garage door 320 is open or closed. The control circuit 334 can request information regarding the current state of the garage door 320. If the control circuit 334 receives a response (e.g., from the garage door opener) indicating that the garage door 320 is open, the control circuit 334 can notify user 308 that the garage door is open, or generate a control signal to cause the garage door opener to close the garage door 320. For example, the control circuit 334 can send a message to user device 310 indicating that the garage door is open. As another example, the control circuit 334 can vibrate the bed 302. As yet another example, the control circuit 334 may generate and transmit a control signal causing the lighting system 314 to flash one or more lights in the bedroom and alert user 308 to check user device 310 for a warning (in this example, a warning about the garage door 320 being open). Alternatively or additionally, the control circuit 334 may generate and transmit a control signal causing the garage door opener to close the garage door 320 in response to the identification that user 308 is in bed at night and that the garage door 320 is open. In some implementations, the control signal may vary depending on the age of user 308.

[0090] The control circuit 334 may also transmit and receive communications to control or receive state information related to the door 332 or the oven 322. For example, if it detects that user 308 is in bed at night, the control circuit 334 may generate and transmit a request to a device or system for detecting the state of door 332. The information returned in response to the request may indicate various states of door 332, such as open, closed but unlocked, or closed and locked. If door 332 is open or closed but unlocked, the control circuit 334 may warn user 308 about the state of the door, for example, as described above for the garage door 320. Alternatively or in addition to the warning to user 308, the control circuit 334 may generate and transmit a control signal to lock or close and lock door 332. If door 332 is closed and locked, the control circuit 334 may determine that no further action is required.

[0091] Similarly, when the control circuit 334 detects that user 308 is in bed at night, it may generate and send a request to oven 322 to request the oven's state (e.g., on or off). If oven 322 is on, the control circuit 334 may warn user 308 and / or generate and send a control signal to turn oven 322 off. If oven is already off, the control circuit 334 may determine that no further action is required. In some implementations, various alerts may be generated for various events. For example, the control circuit 334 may flash lamp 326 (or one or more other lights via the lighting system 314) in a first pattern if the security system 318 detects a breach, in a second pattern if the garage door 320 is open, in a third pattern if door 332 is open, in a fourth pattern if the oven 322 is on, and in a fifth pattern if another bed detects that the user of that bed has woken up (for example, when a sensor in the child's bed 302 detects that user 308's child has gotten out of bed in the middle of the night). Other examples of warnings that can be processed by the control circuit 334 of bed 302 and communicated to the user include a smoke detector that detects smoke (and communicates the detection of such smoke to the control circuit 334), a carbon monoxide tester that detects carbon monoxide, a heater malfunction, or any other device that can communicate with the control circuit 334 and is capable of detecting the occurrence of an event that should draw the attention of user 308.

[0092] The control circuit 334 may also communicate with a system or device for controlling the state of the blinds 330. For example, in response to a determination that user 308 is in bed at night, the control circuit 334 may generate and transmit a control signal to close the blinds 330. As another example, in response to a determination that user 308 has gotten up that day (for example, the user got out of bed after 6:30 a.m.), the control circuit 334 may generate and transmit a control signal to open the blinds 330. In contrast, if user 308 gets out of bed before the user 308's usual wake-up time, the control circuit 334 may determine that user 308 has not yet gotten up that day and will not generate a control signal to open the blinds 330. As yet another example, the control circuit 334 may generate and transmit a control signal to close the first set of blinds in response to the detection of user 308's presence in bed, and to close the second set of blinds in response to the detection that the user is asleep.

[0093] The control circuit 334 may generate and transmit control signals to control the functions of other household appliances in response to detection of user interaction with the bed 302. For example, in response to a determination that user 308 has woken up for the day, the control circuit 334 may generate and transmit a control signal to the coffee maker 324 to cause the coffee maker 324 to start brewing coffee. As another example, the control circuit 334 may generate and transmit a control signal to the oven 322 to cause the oven to start preheating (for users who like freshly baked bread in the morning). As yet another example, the control circuit 334 may use information indicating that user 308 has woken up for the day, along with information indicating that it is currently winter and / or that the outside temperature is below a threshold, to generate and transmit a control signal to turn on the car's engine block heater.

[0094] As another example, the control circuit 334 may generate and transmit control signals to put one or more devices into sleep mode in response to detection of the presence of user 308 in bed or in response to detection that user 308 is asleep. For example, the control circuit 334 may generate a control signal to switch user 308's mobile phone into sleep mode. The control circuit 334 may then transmit this control signal to the mobile phone. Further later, if it is determined that user 308 has woken up that day, the control circuit 334 may generate and transmit a control signal to switch the mobile phone out of sleep mode (to normal mode).

[0095] In some implementations, the control circuit 334 may communicate with one or more noise control devices. For example, if it is determined that user 308 is in bed at night or is asleep, the control circuit 334 may generate and transmit control signals to activate one or more noise cancellation devices. The noise cancellation devices may be included as part of the bed 302, for example, or placed in the bedroom where the bed 302 is located. As another example, if it is determined that user 308 is in bed at night or is asleep, the control circuit 334 may generate and transmit control signals to turn the volume on, off, up, or down for one or more sound-generating devices, such as a stereo system radio, computer, or tablet.

[0096] Furthermore, the functions of the bed 302 are controlled by the control circuit 334 in response to user interaction with the bed 302. For example, the bed 302 may include an adjustable base and an articulated controller configured to adjust the position of one or more parts of the bed 302 by adjusting the adjustable base that supports the bed. For example, the articulated controller may adjust the bed 302 from a flat position to a position in which the head portion of the mattress of the bed 302 is tilted upward (for example, to facilitate the user sitting on the bed and / or watching television). In some implementations, the bed 302 includes several independently articulated sections. For example, parts of the bed corresponding to the positions of air chambers 306a and 306b may be articulated independently of each other to allow one person to rest in a first position (e.g., a flat position) while a second person rests in a second position (e.g., a reclined position with their head tilted diagonally above their waist) while one person rests on the surface of the bed 302. In some implementations, separate positions may be set for two different beds (e.g., two twin beds placed next to each other). The base of the bed 302 may include two or more zones that can be adjusted independently. The joint motion controller may also be configured to provide different levels of massage to one or more users on the bed 302, or to vibrate the bed to alert the user 308 as described above.

[0097] The control circuit 334 may adjust the position (e.g., tilt and lowering positions for user 308 and / or additional users of the bed 302) in response to user interaction with the bed 302. For example, the control circuit 334 may cause the joint motion controller to adjust the bed 302 to a first reclining position for user 308 in response to sensing the presence of user 308 in the bed. The control circuit 334 may cause the joint motion controller to adjust the bed 302 to a second reclining position (e.g., a less reclined or flat position) in response to determining that user 308 is asleep. As another example, the control circuit 334 may receive a communication from the television 312 indicating that user 308 has turned off the television 312, and in response, the control circuit 334 may cause the joint motion controller to adjust the position of the bed 302 to a preferred user sleeping position (e.g., user 308 turning off the television 312 while in bed indicates that user 308 wants to fall asleep).

[0098] In some implementations, the control circuit 334 may control the joint motion controller to wake one user of bed 302 without waking another user of bed 302. For example, user 308 and the second user of bed 302 may each set different wake-up times (e.g., 6:30 AM and 7:15 AM, respectively). When it is user 308's wake-up time, the control circuit 334 may cause the joint motion controller to vibrate only one side of the bed where user 308 is located, or to change its position, thereby waking user 308 without disturbing the second user. When it is the second user's wake-up time, the control circuit 334 may cause the joint motion controller to vibrate only the side of the bed where the second user is located, or to change its position. Alternatively, when it is the second user's wake-up time, the control circuit 334 may use other methods (e.g., an audio alarm or turning on a light) to wake the second user. This is because when the control circuit 334 attempts to wake up the second user, user 308 is already awake and therefore cannot be disturbed.

[0099] Continuing to refer to Figure 3, the control circuit 334 of bed 302 can use information about the interactions between multiple users with bed 302 to generate control signals to control the functions of various other devices. For example, the control circuit 334 may wait to generate control signals to activate, for example, the security system 318 or to instruct the lighting system 314 to turn off the lights in various rooms, until it is detected that both user 308 and the second user are present on bed 302. As another example, upon detecting the presence of user 308 on bed, the control circuit 334 may generate a first set of control signals to turn off a first set of lights in the lighting system 314, and in response to detecting the presence of the second user on bed, it may generate a second set of control signals to turn off a second set of lights. As yet another example, the control circuit 334 may wait to generate a control signal to open the blinds 330 until it is determined that both user 308 and the second user have woken up for the day. As yet another example, in response to the determination that user 308 is awake and out of bed that day, but second user is still asleep, the control circuit 334 may generate and transmit a first set of control signals, causing the coffee maker 324 to start brewing coffee, the security system 318 to be deactivated, the lamp 326 to be turned on, the night light 328 to be turned off, the thermostat 316 to raise the temperature of one or more rooms to 72 degrees Fahrenheit, and the blinds (e.g., blinds 330) in rooms other than the bedroom where the bed 302 is located. Subsequently, in response to the detection that second user is no longer in bed (or second user is awake), the control circuit 334 may generate and transmit a second set of control signals, for example, causing the lighting system 314 to turn on one or more lights in the bedroom, open the bedroom blinds, and turn on the television 312 to a predetermined channel.

[0100] [Example of a data processing system associated with a bed]

[0101] Here, examples of systems and components that may be used for data processing tasks associated with a bed are described. In some cases, multiple examples of a particular component or group of components are presented. Some of these examples are redundant and / or mutually exclusive substitutes. Connections between components are shown as examples illustrating possible network configurations for allowing communication between components. Various forms of connections may be used as technically required or desired. Such connections generally represent logical connections that can be created in any technically feasible manner. For example, a network on a motherboard may be created with printed circuit boards, wireless data connections, and / or other types of network connections. Some logical connections are not illustrated for clarity. For example, many or all elements of a particular component may need to be connected to power and / or computer-readable memory, but for clarity, connections to power and / or computer-readable memory may not be illustrated.

[0102] Figure 4A is a block diagram of an example of a data processing system 400 that may be associated with a bed system. It includes those described above with respect to Figures 1 to 3. The system 400 includes a pump motherboard 402 and a pump daughterboard 404. The system 400 also includes a sensor array 406, which may include one or more sensors configured to sense environmental and / or physical phenomena of the bed and report such senses to the pump motherboard 402, for example, for analysis. The system 400 also includes a controller array 408, which may include one or more controllers configured to control logical control devices of the bed and / or environment. The pump motherboard 400 may communicate with one or more computing devices 414 and one or more cloud services 410 via a local network, via the internet 412, or via other technically appropriate means. Each of these components will be described in more detail below, along with several exemplary forms.

[0103] In this example, the pump motherboard 402 and the pump daughterboard 404 are connected in a communicative manner. They can be conceptually described as the center or hub of system 400, and the other components can be conceptually described as spokes of system 400. In some forms, this may mean that each of the spoke components communicates primarily or exclusively with the pump motherboard 402. For example, the sensors of a sensor array may not be configured to communicate directly with their corresponding controllers, or may not be able to. Instead, each spoke component can communicate with the motherboard 402. The sensors of sensor array 406 may report sensor readings to the motherboard 402, which in turn may determine whether the controllers of controller array 408 should adjust some parameters of the logic control devices or correct the state of one or more peripheral devices. If, in some case, the bed temperature is determined to be too high, the pump motherboard 402 may determine that the temperature controller should cool the bed.

[0104] One advantage of a hub-spoke network configuration (sometimes called a star network) is reduced network traffic compared to, for example, a mesh network using dynamic routing. Even if a particular sensor generates a large, continuous stream of traffic, that traffic may only be sent to the motherboard 402 via one spoke of the network. The motherboard 402 may, for example, marshal the data, condense it into a smaller data format, and retransmit it for storage in the cloud service 410. Additionally or alternatively, the motherboard 402 may, in response to a large stream, generate a single small command message to be sent via different spokes of the network. For example, if the large data stream is pressure readings sent several times per second from the sensor array 406, the motherboard 402 may respond to the controller array with a single command message to increase the pressure in the air chamber. In this case, the single command message may be orders of magnitude smaller than the stream of pressure readings.

[0105] Another advantage is that the hub-spoke network configuration can accommodate a scalable network that can handle the addition, removal, and failure of components. This could, for example, allow for more, fewer, or different sensors in the sensor array 406, more, fewer, or different controllers in the controller array 408, more, fewer, or different computing devices 414, and / or more, fewer, or different cloud services 410. For example, if a particular sensor fails or is superseded by a newer version of that sensor, the system 400 could be configured such that only the motherboard 402 needs to be updated for the replacement sensor. This could allow for product differentiation, for example, that the same motherboard 402 can support entry-level products with fewer sensors and controllers, higher-value products with more sensors and controllers, and customer personalization where customers can add their own selected components to the system 400.

[0106] Furthermore, a series of airbed products may utilize system 400 with various components. In applications where all airbeds in the product line include both a central logic unit and a pump, the motherboard 402 (and optionally the daughterboard 404) can be designed to fit within a single universal housing. Subsequently, with each product upgrade within the product line, additional sensors, controllers, cloud services, etc., can be added. Designing all products in the product line from such a base can reduce design, manufacturing, and testing times compared to a product line where each product has a custom logic control system.

[0107] Each of the aforementioned components can be implemented in various technologies and forms. Several examples of each component are described below. In some alternatives, two or more components of system 400 may be implemented by a single alternative component, some components may be implemented by multiple separate components, and / or some functions may be provided by different components.

[0108] Figure 4B is a block diagram showing several communication paths for the data processing system 400. As previously mentioned, the motherboard 402 and the pump daughterboard 404 can function as hubs for peripheral devices and cloud services of the system 400. When the pump daughterboard 404 communicates with cloud services or other components, the communication from the pump daughterboard 404 can be routed through the pump motherboard 402. This allows, for example, the bed to have only a single connection to the internet 412. The computing device 414 may also have a connection to the internet 412, sometimes via the same gateway used by the bed and / or sometimes via a different gateway (e.g., a cell service provider).

[0109] Several cloud services 410 have been described previously. As shown in Figure 4B, some cloud services, such as cloud services 4l0d and 4l0e, can be configured so that the pump motherboard 402 can communicate directly with them. That is, the motherboard 402 can communicate with the cloud services 410 without needing to use another cloud service 410 as an intermediary. Additionally or alternatively, some cloud services 410, such as cloud service 410f, may be reachable by the pump motherboard 402 only through an intermediary cloud service, such as cloud service 410e. Although not shown here, some cloud services 410 may be reachable by the pump motherboard 402 directly or indirectly.

[0110] Furthermore, some or all of the cloud services 410 may be configured to communicate with other cloud services. This communication may include the transfer of data and / or remote function calls in any technically appropriate format. For example, one cloud service 410 may request a copy of 410 data from another cloud service, for example, for backup, coordination, migration purposes, or for performing computation or data mining. In another example, many cloud services 410 may contain data indexed according to specific users tracked by user account clouds 410c and / or bed data clouds 410a. These cloud services 410 may communicate with the user account clouds 410c and / or bed data clouds 410a when accessing data specific to a particular user or bed.

[0111] Figure 5 is a block diagram of an example of a motherboard 402 that may be used in a data processing system that may be associated with a bed system, including those described above in relation to Figures 1 to 3. In this example, the motherboard 402 may be limited to having relatively few components and providing a relatively limited set of features compared to other examples described below.

[0112] The motherboard includes a power supply unit 500, a processor 502, and computer memory 512. Generally, the power supply unit includes hardware used to receive power from an external power source and supply it to the components of the motherboard 402. The power supply unit may include, for example, a battery pack and / or a wall outlet adapter (plug), an AC-DC converter, a DC-AC converter, a power regulator, a capacitor bank, and / or one or more interfaces for providing power in terms of current type, voltage, etc., as required by the other components of the motherboard 402.

[0113] A processor 502 is generally a device for receiving inputs, performing logical decisions, and providing outputs. A processor 502 can be a central processing unit, a microprocessor, a general-purpose logic circuit, an application-specific integrated circuit (ASIC), a combination of these, and / or other hardware for performing the required functions.

[0114] Memory 512 is generally one or more devices for storing data. Memory 512 may include long-term stable data storage (e.g., on a hard disk), short-term unstable data storage (e.g., on random-access memory), or any other technically appropriate configuration.

[0115] The motherboard 402 includes a pump controller 504 and a pump motor 506. The pump controller 504 may receive commands from the processor 502 and, in response, control the function of the pump motor 506. For example, the pump controller 504 may receive a command from the processor 502 to increase the pressure in the air chamber by 0.3 pounds per square inch (PSI). In response, the pump controller 504 may actuate a valve so that the pump motor 506 is configured to pump air into the selected air chamber and operate the pump motor 506 for a time corresponding to 0.3 PSI, or until a sensor indicates that the pressure has increased by 0.3 PSI. In an alternative form, the message may specify that the chamber should be inflated to a target PSI, and the pump controller 504 may operate the pump motor 506 until that target PSI is reached.

[0116] The valve solenoid 508 can control which air chamber the pump is connected to. In some cases, the solenoid 508 may be directly controlled by the processor 502. In some cases, the solenoid 508 may be controlled by the pump controller 504.

[0117] The remote interface 510 of the motherboard 402 may allow the motherboard 402 to communicate with other components of the data processing system. For example, the motherboard 402 may be able to communicate with one or more daughterboards, peripheral sensors, and / or peripheral controllers via the remote interface 510. The remote interface 510 may provide any technically appropriate communication interface, including but not limited to multiple communication interfaces such as WiFi, Bluetooth®, and copper wired networks.

[0118] Figure 6 is a block diagram of an example of a motherboard 402 that may be used in a data processing system that may be associated with a bed system, including those described above with reference to Figures 1 to 3. Compared to the motherboard 402 described with reference to Figure 5, the motherboard in Figure 6 may contain more components and may offer more functionality in some applications.

[0119] In addition to the power supply unit 500, processor 502, pump controller 504, pump motor 506, and valve solenoid 508, the motherboard 402 is shown together with a valve controller 600, pressure sensor 602, universal serial bus (USB) stack 604, WiFi radio 606, Bluetooth® Low Energy (BLE) radio 608, ZigBee radio 610, Bluetooth® radio 612, and computer memory 512.

[0120] Similar to how the pump controller 504 translates commands from the processor 502 into control signals for the pump motor 506, the valve controller 600 may translate commands from the processor 502 into control signals for the valve solenoid 508. For example, the processor 502 may issue a command to the valve controller 600 to connect the pump to a specific air chamber among a group of air chambers in an airbed. The valve controller 600 may control the position of the valve solenoid 508 so that the pump is connected to the indicated air chamber.

[0121] The pressure sensor 602 can read pressure readings from one or more air chambers of the airbed. The pressure sensor 602 can also perform digital sensor calibration.

[0122] Motherboard 402 may include a set of network interfaces, including but not limited to those illustrated herein. These network interfaces may allow the motherboard to communicate with any number of devices, including but not limited to peripheral sensors, peripheral controllers, computing devices, and devices and services connected to the Internet 412, via a wired or wireless network.

[0123] Figure 7 is a block diagram of an example of a daughterboard 404 that may be used in a data processing system that may be associated with a bed system, including those described above with respect to Figures 1 to 3. In some forms, one or more daughterboards 404 may be connected to a motherboard 402. Some daughterboards 404 may be designed to offload specific tasks and / or partitioned tasks from the motherboard 402. This may be advantageous, for example, if a particular task is computationally intensive, proprietary, or subject to future revisions. For example, a daughterboard 404 may be used to calculate a specific sleep data metric. This metric may be computationally intensive, and calculating the sleep metric on the daughterboard 404 may free up resources on the motherboard 402 while the metric is being calculated. Additionally and / or alternatively, the sleep metric may be subject to future revisions. It is possible that to update system 400 with a new sleep metric, only the daughterboard 404 that calculates the metric needs to be replaced. In this case, since the same motherboard 402 and other components may be used, it is not necessary to perform unit testing on additional components as well as daughterboard 404.

[0124] The daughterboard 404 is illustrated together with a power supply 700, a processor 702, computer-readable memory 704, a pressure sensor 706, and a WiFi radio 708. The processor may use the pressure sensor 706 to collect information about the pressure in one or more air chambers of the airbed. From this data, the processor 702 may execute an algorithm to calculate a sleep metric. In some examples, the sleep metric may be calculated from the air chamber pressure alone. In other examples, the sleep metric may be calculated from one or more other sensors. In examples where various data are required, the processor 702 may receive such data from appropriate one or more sensors. These sensors may be located inside the daughterboard 404, may be accessible via the WiFi radio 708, or may be communicating with the processor 702. Once the sleep metric is calculated, the processor 702 may report the sleep metric to, for example, the motherboard 402.

[0125] Figure 8 is a block diagram of an example of a motherboard 800 without a daughterboard, which may be used in a data processing system that may be associated with a bed system, including those described above with respect to Figures 1 to 3. In this example, the motherboard 800 may perform almost all, all, or more of the functions described with reference to the motherboard 402 in Figure 6 and the daughterboard 404 in Figure 7.

[0126] Figure 9 is a block diagram of an example of a sensor array 406 that may be used in a data processing system that may be associated with a bed system, including those described above with respect to Figures 1 to 3. Generally, the sensor array 406 is a conceptual group of some or all of the peripheral sensors that communicate with the motherboard 402 but are not native to the motherboard 402.

[0127] Peripheral sensors of the sensor array 406 may communicate with the motherboard 402 via one or more network interfaces of the motherboard, including but not limited to a USB stack 604, a WiFi radio 606, a Bluetooth® Low Energy (BLE) radio 608, a ZigBee radio 610, and a Bluetooth® radio 612, as appropriate for the specific sensor configuration. For example, a sensor that outputs readings via a USB cable may communicate via the USB stack 604.

[0128] Some of the peripheral sensors 900 of the sensor array 406 may be mounted on the bed. These sensors may, for example, be embedded within the bed structure and sold with the bed, or may be mounted on the bed structure at a later date. Other peripheral sensors 902, 904 may communicate with the motherboard 402 but may not be selectively mounted on the bed. In some cases, some or all of the bed-mounted sensors 900 and / or peripheral sensors 902, 904 may share networking hardware. This includes conductors (wires) from each sensor, including wires, multiwire cables, or plugs, which connect all the associated sensors to the motherboard 402 when mounted on the motherboard 402. In some embodiments, one, some, or all of the sensors 902, 904, 906, 908, 910 may be capable of sensing one or more features of the mattress, such as pressure, temperature, light, sound, and / or one or more other features of the mattress. In some embodiments, one, some, or all of the sensors 902, 904, 906, 908, and 910 are capable of sensing one or more external features of the mattress. In some embodiments, while some or all of the sensors 902, 904, 906, 908, and 910 are capable of sensing one or more features of the mattress and / or one or more external features of the mattress, the pressure sensor 902 is capable of sensing the pressure of the mattress.

[0129] Figure 10 is a block diagram of an example controller array 408 that may be used in a data processing system that may be associated with a bed system, including those described above with respect to Figures 1 to 3. Generally, the controller array 408 is a conceptual group of some or all peripheral controllers that communicate with the motherboard 402 but are not native to the motherboard 402.

[0130] Peripheral controllers of the controller array 408 may communicate with the motherboard 402 via one or more network interfaces of the motherboard, including but not limited to a USB stack 604, a WiFi radio 606, a Bluetooth® Low Energy (BLE) radio 608, a ZigBee radio 610, and a Bluetooth® radio 612, as appropriate for a particular sensor configuration. For example, a controller that receives commands via a USB cable may communicate via the USB stack 604.

[0131] Some of the controllers 1000 of the controller array 408 may be mounted on the bed. This includes, but is not limited to, the temperature controller 1006, the lighting controller 1008, and / or the speaker controller 1010. These controllers may, for example, be embedded within the bed structure and sold with the bed, or may be mounted on the bed structure later. Other peripheral controllers 1002, 1004 may communicate with the motherboard 402 but may not be selectively mounted on the bed. In some cases, some or all of the bed-mounted controllers 1000 and / or peripheral controllers 1002, 1004 may share networking hardware. This includes conductors (wires) including wires, multi-wire cables, or plugs for each controller that connect all the relevant controllers to the motherboard 402 when mounted on the motherboard 402.

[0132] Figure 11 is a block diagram of an example of a computing device 414 that may be used in a data processing system that may be associated with a bed system, including those described above with respect to Figures 1 to 3. The computing device 414 may include, for example, a computing device used by the bed user. The exemplary computing device 414 includes, but is not limited to, mobile computing devices (e.g., mobile phones, tablet computers, laptops) and desktop computers.

[0133] The computing device 414 includes a power supply unit 1100, a processor 1102, and computer-readable memory 1104. User inputs and outputs may be transmitted by, for example, a speaker 1106, a touchscreen 1108, or other components not shown, such as a pointing device or keyboard. The computing device 414 may run one or more applications 1110. These applications may include, for example, applications that allow a user to interact with the system 400. These applications may allow the user to view information about the bed (such as sensor readings and sleep metrics) or configure the operation of the system 400 (for example, setting a desired firmness for the bed or setting a desired operation for peripheral devices). In some cases, the computing device 414 may be used in addition to, or to replace, the remote control 122 described above.

[0134] Figure 12 is a block diagram of an example of a bed data cloud service 410a that may be used in a data processing system that may be associated with a bed system, including those described above with respect to Figures 1 to 3. In this example, the bed data cloud service 410a is configured to collect sensor data and sleep data from a specific bed and to match the sensor data and sleep data with one or more users who were using the bed when the sensor data and sleep data were generated.

[0135] The bed data cloud service 410a is shown together with a network interface 1200, a communication manager 1202, server hardware 1204, and server system software 1206. Furthermore, the bed data cloud service 410a is shown together with a user identification module 1208, a device management module 1210, a sensor data module 1212, and an advanced sleep data module 1214.

[0136] The network interface 1200 generally includes hardware and low-level software used to allow one or more hardware devices to communicate over a network. For example, the network interface 1200 may include network cards, routers, modems, and other hardware required to allow components of the bed data cloud service 410a to communicate with each other and with other destinations, for example, over the internet 412. The communication manager 1202 generally includes hardware and software operating on the network interface 1200. This includes software for initiating, maintaining, and ending network communications used by the bed data cloud service 410a. This includes, for example, TCP / IP, SSL or TLS, torrent, and other communication sessions over local or wide area networks. The communication manager 1202 may also provide load balancing and other services to other elements of the bed data cloud service 410a.

[0137] Server hardware 1204 generally includes physical processing units used to instantiate and maintain the bed data cloud service 410a. This hardware includes, but is not limited to, processors (e.g., central processing units, ASICs, graphics processors) and computer-readable memory (e.g., random access memory, stable hard disks, tape backups). One or more servers may be configured in a cluster, multicomputer, or data center, which may be geographically separated or connected.

[0138] The server system software 1206 generally includes software that runs on the server hardware 1204 to provide an operating environment for applications and services. The server system software 1206 may include an operating system that runs on the physical server, virtual machines that are instantiated on the physical server to generate many virtual servers, and server-level operations such as data migration, redundancy, and backup.

[0139] The user identification module 1208 may include or refer to data relating to users of beds equipped with associated data processing systems. For example, a user may include a customer, owner, or other user registered with the bed data cloud service 410a or other services. Each user may have, for example, a unique identifier, user credentials, contact information, billing information, demographic information, or other technically appropriate information.

[0140] The device management module 1210 may include or reference data related to beds or other products associated with the data processing system. For example, a bed may include product information (product information) sold or registered in the system associated with the bed data cloud service 410a. Each bed may have, for example, a unique identifier, model and / or serial number, sales information, geographical information, delivery information, a list of associated sensors and peripheral control devices, etc. Furthermore, one or more indexes stored by the bed data cloud service 410a may identify users associated with a bed. For example, this index may record sales of a bed to one or more users sleeping in the bed.

[0141] The sensor data module 1212 may record raw (unprocessed) or compressed (processed) sensor data recorded by a bed equipped with an associated data processing system. For example, the bed's data processing system may include a temperature sensor, a pressure sensor, and a light sensor. Readings from these sensors, either in their raw sensor form or in a format generated from the raw data (e.g., sleep metrics), may be communicated by the bed's data processing system to the bed data cloud service 410a and stored in the sensor data module 1212. Furthermore, one or more indices stored by the bed data cloud service 410a may identify the user and / or bed associated with the sensor data module 1212.

[0142] The bed data cloud service 410a may generate advanced sleep data 1214 using any of its available data. Generally, advanced sleep data 1214 includes sleep metrics and other data generated from sensor readings. Some of these calculations may be performed in the bed data cloud service 410a instead of being performed locally in the bed data processing system, for example, if the calculations are complex or require a large amount of memory or processor power that is not available in the bed data processing system. This can be useful in allowing the bed system to operate with a relatively simple controller while still being part of a system that performs relatively complex tasks and calculations.

[0143] Figure 13 is a block diagram of an example of a sleep data cloud service 410b that may be used in a data processing system that may be associated with a bed system, including those described above in relation to Figures 1 to 3. In this example, the sleep data cloud service 410b is configured to record data related to the user's sleep experience.

[0144] The sleep data cloud service 410b is shown together with a network interface 1300, a communication manager 1302, server hardware 1304, and server system software 1306. Furthermore, the sleep data cloud service 410b is shown together with a user identification module 1308, a pressure sensor management module 1310, a pressure-based sleep data module 1312, a raw pressure sensor data module 1314, and a non-pressure sleep data module 1316.

[0145] The pressure sensor management module 1310 may include or refer to data related to the configuration and operation of pressure sensors in the bed. For example, this data may include identifiers of the sensor type for a particular bed, their settings and calibration data, and so on.

[0146] The pressure-based sleep data 1312 can be used with the raw pressure sensor data 1314 to calculate sleep metrics, particularly those associated with the pressure sensor data. For example, the user's presence, movement, weight change, heart rate, and respiratory rate can all be determined from the raw pressure sensor data 1314. Furthermore, one or more indices stored by the sleep data cloud service 410b can identify the user associated with the pressure sensor, the raw pressure sensor data, and / or the pressure-based sleep data.

[0147] Non-pressure sleep data 1316 can be used to calculate sleep metrics using other data sources. For example, user-entered preferences, light sensor readings, and acoustic sensor readings can all be used to track sleep data. Furthermore, one or more indices stored by the sleep data cloud service 410b can identify users associated with other sensors and / or non-pressure sleep data 1316.

[0148] Figure 14 is a block diagram of an example of a user account cloud service 410c that may be used in a data processing system that may be associated with a bed system, including those described above with respect to Figures 1 to 3. In this example, the user account cloud service 4l0c is configured to record a list of users and identify other data associated with those users.

[0149] The user account cloud service 410c is shown together with the network interface 1400, the communication manager 1402, the server hardware 1404, and the server system software 1406. Furthermore, the user account cloud service 410c is shown together with the user identification module 1408, the purchase history module 1410, the engagement module 1412, and the application usage history module 1414.

[0150] The user identification module 1408 may include or refer to data relating to a user of a bed equipped with an associated data processing system. For example, a user may include a customer, owner, or other user registered with the user account cloud service 410c or other services. Each user may have, for example, a unique identifier, user credentials, demographic information, or other technically appropriate information.

[0151] The purchase history module 1410 may contain or reference data related to user purchases. For example, purchase data may include sales contact information, invoice information, and sales representative information. Furthermore, one or more indexes stored by the user account cloud service 410c may identify the user associated with the purchase.

[0152] The engagement module 1412 may track user interactions with manufacturers, vendors, and / or administrators of beds and / or cloud services. This engagement data may include communications (e.g., emails, service calls, etc.), sales data (e.g., receipts, configuration logs), and social network interactions.

[0153] The usage history module 1414 may contain data on user interactions with one or more applications and / or remote controls of the bed. For example, a monitoring and configuration application may be distributed to run on multiple computing devices 412, for example. This application may log and report user interactions for storage in the application usage history module 1414. Furthermore, one or more indices stored by the user account cloud service 410c may identify the user associated with each log entry.

[0154] Figure 15 is a block diagram of an example of a point-of-sale (POS) cloud service 1500 that may be used in a data processing system that may be associated with a bed system, including those described above in relation to Figures 1 to 3. In this example, the point-of-sale cloud service 1500 is configured to record data related to user purchases.

[0155] The point-of-sale (POS) cloud service 1500 is shown together with a network interface 1502, a communications manager 1504, server hardware 1506, and server system software 1508. Furthermore, the point-of-sale (POS) cloud service 1500 is shown together with a user identification module 1510, a purchase history module 1512, and a setup module 1514.

[0156] The purchase history module 1512 may include, or refer to, data related to purchases made by a user identified by the user identification module 1510. Purchase information may include, for example, data on the sale, price, place of sale, delivery address, and configuration options selected by the user at the time of sale. These configuration options may include the user's choices regarding how they want the newly purchased bed to be set up, and may also include, for example, an expected sleep schedule, a list of peripheral sensors and controllers that the user has or will install, etc.

[0157] The bed setup module 1514 may include or refer to data related to the setup of a bed purchased by the user. The bed setup data may include, for example, the date and address on which the bed will be delivered, the person receiving the delivery, the configuration applied to the bed at the time of delivery, the names of one or more people who will sleep on the bed, and which side of the bed each person will use.

[0158] Data recorded in the point-of-sale (POS) cloud service 1500 can later be referenced by the user's bed system to control the functions of the bed system and / or send control signals to peripheral components according to the data recorded in the PSOS cloud service 1500. This allows salespeople to collect information from users at the point of sale, which facilitates the later automation of the bed system. In some examples, some or all features of the bed system can be automated, and little to no user input data is required after the point of sale. In other examples, data recorded in the PSOS cloud service 1500 can be used in relation to various additional data collected from user input data.

[0159] Figure 16 is a block diagram of an example of an environmental cloud service 1600 that may be used in a data processing system that may be associated with a bed system, including those described above in relation to Figures 1 to 3. In this example, the environmental cloud service 1600 is configured to record data related to the user's home environment.

[0160] The environmental cloud service 1600 is shown together with a network interface 1602, a communication manager 1604, server hardware 1606, and server system software 1608. Furthermore, the environmental cloud service 1600 is shown together with a user identification module 1610, an environmental sensor module 1612, and an environmental factor module 1614.

[0161] The environmental sensor module 1612 may include a list of sensors installed in the bed by the user of the user identification module 1610. These sensors may include any sensors capable of detecting environment variables, such as light sensors, noise sensors, vibration sensors, and thermostats. Furthermore, the environmental sensor module 1612 may store past readings or reports from those sensors.

[0162] The environmental factors module 1614 may include reports generated based on data from the environmental sensor module 1612. For example, for a user with a light sensor, the environmental factors module 1614 may hold reports indicating the frequency and duration of instances of increased lighting while the user is sleeping.

[0163] In the examples described herein, each cloud service 410 is shown with some identical components. In various forms, these identical components may be shared partially or completely between services, or they may be separate. In some forms, each service may have separate copies of some or all of the components that are the same or different in some respects. Furthermore, these components are provided only as illustrative examples. In other examples, each cloud service may have a different number, type, and style of components, depending on what is technically possible.

[0164] Figure 17 is a block diagram of an example of automating peripheral devices around a bed using a data processing system that may be associated with the bed (such as a bed in the bed system described herein). Shown here is a behavior analysis module 1700 running on the pump motherboard 402. For example, the behavior analysis module 1700 may be one or more software components stored in computer memory 512 and executed by processor 502. Generally, the behavior analysis module 1700 may collect data from a wide variety of sources (e.g., sensors, non-sensor local sources, cloud data services) and use behavior algorithms 1702 to generate one or more actions to be taken (e.g., commands to send to peripheral controllers, data to send to cloud services). This may be useful, for example, for tracking user behavior or automating devices that communicate with the user's bed.

[0165] The behavioral analysis module 1700 may collect data from any technically appropriate source to collect data about, for example, the bed's characteristics, the bed's environment, and / or the bed's user. Some such sources include any of the sensors in the sensor array 406. For example, this data may provide the behavioral analysis module 1700 with information about the current state of the environment surrounding the bed. For example, the behavioral analysis module 1700 may access readings from the pressure sensor 902 to determine the pressure in the air chamber within the bed. From these readings, and possibly other data, the presence of a user in the bed may be determined. In another example, the behavioral analysis module 1700 may access the light sensor 908 to detect the amount of light in the bed's environment.

[0166] Similarly, the behavioral analysis module 1700 may access data from cloud services. For example, the behavioral analysis module 1700 may access the bed cloud service 410a and access historical sensor data 1212 and / or advanced sleep data 1214. Other cloud services 410, including those not previously described, may be accessed by the behavioral analysis module 1700. For example, the behavioral analysis module 1700 may access weather reporting services, third-party data providers (e.g., traffic and news data, emergency broadcast data, user travel data), and / or clock and calendar services.

[0167] Similarly, the behavioral analysis module 1700 may access data from non-sensor sources 1704. For example, the behavioral analysis module 1700 may access local clock and calendar services (e.g., components of the motherboard 402 or processor 502).

[0168] The behavior analysis module 1700 can aggregate and prepare this data for use by one or more behavior algorithms 1702. The behavior algorithms 1702 can be used to learn user behavior and / or to perform certain actions based on the state of the accessed data and / or predicted user behavior. For example, the behavior algorithm 1702 can use available data (e.g., pressure sensor, non-sensor data, clock and calendar data) to create a model of when the user goes to sleep each night. The same or different behavior algorithms 1702 can then be used to determine whether an increase in air chamber pressure is likely to indicate the user is going to sleep, and if so, some data can be sent to a third-party cloud service 410 and / or to activate devices such as the pump controller 504, base actuator 1706, temperature controller 1008, under-bed lighting 1010, peripheral controller 1002, or peripheral controller 1004.

[0169] In the illustrated example, the behavioral analysis module 1700 and the behavioral algorithm 1702 are shown as components of the motherboard 402. However, other configurations are possible. For example, the same or similar behavioral analysis module and / or behavioral algorithm may run on one or more cloud services, and the resulting output may be transmitted to the motherboard 402, the controller of the controller array 408, or any other technically appropriate recipient.

[0170] Figure 18 shows an example of a computing device 1800 and an example of a mobile computing device that may be used to implement the techniques described herein. The computing device 1800 is intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal information terminals, servers, blade servers, mainframes, and other suitable computers. The mobile computing device is intended to represent various forms of mobile devices, such as personal information terminals, mobile phones, smartphones, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are for illustrative purposes only and are not intended to limit the implementation of the invention described and / or claimed herein.

[0171] The computing device 1800 includes a processor 1802, memory 1804, storage device 1806, a high-speed interface 1808 connected to memory 1804 and several high-speed expansion ports 1810, and a low-speed interface 1812 connected to a low-speed expansion port 1814 and storage device 1806. Each of the processor 1802, memory 1804, storage device 1806, high-speed interface 1808, high-speed expansion ports 1810, and low-speed interface 1812 can be interconnected using various buses and mounted on a common motherboard or in other forms as needed. The processor 1802 processes instructions for execution within the computing device 1800, including instructions stored in memory 1804 or storage device 1806, and can display graphic information for a GUI on an external input / output device such as a display 1816 coupled to the high-speed interface 1808. In other implementations, multiple processors and / or multiple buses may be used, along with multiple memories and memory types, as needed. Furthermore, multiple computing devices may be connected so that each computing device provides some of the necessary functionality (e.g., as a server bank, a group of blade servers, or a multiprocessor system).

[0172] Memory 1804 stores information within the computing device 1800. In some implementations, memory 1804 is one or more volatile memory units. In some implementations, memory 1804 is one or more non-volatile memory units. Memory 1804 may also be another form of computer-readable medium, such as a magnetic disk or an optical disk.

[0173] The storage device 1806 can provide large-capacity storage to the computing device 1800. In some implementations, the storage device 1806 may be or include computer-readable media such as floppy disk drives, hard disk drives, optical disk drives, tape drives, flash memory, or other similar solid-state memory devices, or an array of multiple devices including a storage area network or other forms of multiple devices. The computer program product may be tangibly embodied within the information carrier. The computer program product may also include instructions that, when executed, perform one or more methods such as those described above. The computer program product may also be tangibly embodied within computer-readable media or machine-readable media such as memory 1804, storage device 1806, or memory on the processor 1802.

[0174] The high-speed interface 1808 manages bandwidth-intensive operation for the computing device 1800, while the low-speed interface 1812 manages lower bandwidth-intensive operation. Such function assignments are merely illustrative. In some implementations, the high-speed interface 1808 is coupled to memory 1804, a display 1816 (e.g., via a graphics processor or accelerator), and a high-speed expansion port 1810 that can accept various expansion cards (not shown). In such implementations, the low-speed interface 1812 is coupled to storage device 1806 and the low-speed expansion port 1814. The low-speed expansion port 1814 may include various communication ports (e.g., USB, Bluetooth®, Ethernet, Wireless Ethernet) and may be coupled to one or more input / output devices such as a keyboard, pointing device, scanner, or network devices such as switches or routers via a network adapter.

[0175] The computing device 1800 can be implemented in several different forms, as shown in the figure. For example, it can be implemented as a standard server 1820, or multiple times in a group of such servers. Furthermore, it can be implemented in a personal computer such as a laptop computer 1822. It can also be implemented as part of a rack server system 1824. Alternatively, components from the computing device 1800 can be combined with other components of a mobile device (not shown), such as a mobile computing device 1850. Each of such devices may include one or more of the computing device 1800 and the mobile computing device 1850, and the entire system may consist of multiple computing devices communicating with each other.

[0176] The mobile computing device 1850 includes, among other things, an input / output device such as a processor 1852, memory 1864, and display 1854, a communication interface 1866, and a transceiver 1868. The mobile computing device 1850 may also be provided with storage devices such as a microdrive or other devices to provide additional storage (recording devices). Each of the processor 1852, memory 1864, display 1854, communication interface 1866, and transceiver 1868 is interconnected using various buses, and some of these components may be mounted on a common motherboard or in other manner as needed.

[0177] The processor 1852 can execute instructions in the mobile computing device 1850, including instructions stored in memory 1864. The processor 1852 may be implemented as a chipset of a chip including several separate analog and digital processors. The processor 1852 may provide coordination of other components of the mobile computing device 1850, such as control of the user interface, applications run by the mobile computing device 1850, and wireless communication by the mobile computing device 1850.

[0178] The processor 1852 can communicate with the user via a control interface 1858 and a display interface 1856 coupled to the display 1854. The display 1854 may be, for example, a TFT display (thin-film transistor liquid crystal display), an OLED (organic light-emitting diode) display, or other suitable display technology. The display interface 1856 may have suitable circuitry for driving the display 1854 to present graphic information and other information to the user. The control interface 1858 may receive commands from the user and translate them for presentation to the processor 1852. Furthermore, an external interface 1862 may provide communication with the processor 1852 and enable short-range communication of the mobile computing device 1850 with other devices. The external interface 1862 may provide, for example, wired communication in some implementations or wireless communication in other implementations, and multiple interfaces may be used.

[0179] Memory 1864 stores information within the mobile computing device 1850. Memory 1864 may be implemented as one or more computer-readable media, one or more volatile memory units, or one or more non-volatile memory units. Extended memory 1874 may also be provided and connected to the mobile computing device 1850 via an extended interface 1872, which may include, for example, a SIMM (Single In-Line Memory Module) card interface. Extended memory 1874 may provide additional storage space for the mobile computing device 1850, or it may store applications or other information for the mobile computing device 1850. Specifically, extended memory 1874 may include instructions for executing or supplementing the aforementioned processes, and may also include security information. Thus, for example, extended memory 1874 may be provided as a security module for the mobile computing device 1850 and may be programmed with instructions that allow the secure use of the mobile computing device 1850. Furthermore, secure applications can be provided via the SIMM card, along with additional information such as identification information placed on the SIMM card in a manner that prevents hacking.

[0180] The memory may include, for example, flash memory and / or NVRAM memory (non-volatile random access memory), as described below. In some implementations, the computer program product is tangibly embodied within an information carrier. The computer program product includes instructions that, when executed, perform one or more methods as described above. The computer program product may be a computer-readable or machine-readable medium, such as memory 1864, extended memory 1874, or memory on processor 1852. In some implementations, the computer program product may be received as a propagated signal, for example, via transceiver 1868 or external interface 1862.

[0181] The mobile computing device 1850 can communicate wirelessly via a communication interface 1866. The communication interface 1866 may optionally include digital signal processing circuitry. The communication interface 1866 can provide communication under various modes or protocols, including, in particular, GSM (Global System for Mobile Communications), SMS (Short Message Service), EMS (Extended Messaging Service), MMS (Multimedia Messaging Service), CDMA (Code Division Multiple Access), TDMA (Time Division Multiple Access), PDC (Personal Digital Cellular), WCDMA (Wideband Code Division Multiple Access), CDMA2000, or GPRS (General-Purpose Packet Radio Service). Such communication may occur, for example, via a transceiver 1868 using radio frequencies. Furthermore, short-range communication may occur using Bluetooth (Registered Trademark), WiFi, or other such transceivers (not shown). Furthermore, the GPS (Global Positioning System) receiver module 1870 may provide additional navigation and location-related radio data to the mobile computing device 1850. This data can be appropriately used by applications running on the mobile computing device 1850.

[0182] The mobile computing device 1850 may also communicate audibly using the audio codec 1860. The audio codec 1860 may receive spoken information from the user and convert it into usable digital information. Similarly, the audio codec 1860 may generate audible sound for the user, for example, through a speaker in the handset of the mobile computing device 1850. Such sound may include sounds from voice calls, recorded sounds (e.g., voice messages, music files, etc.), and sounds generated by applications running on the mobile computing device 1850.

[0183] The mobile computing device 1850 can be implemented in several different forms, as shown in the figure. For example, it can be implemented as a mobile phone 1880. Alternatively, it can be implemented as part of a smartphone 1882, a personal information terminal, or other similar mobile device.

[0184] Various implementations of the systems and technologies described herein can be realized in digital electronic circuits, integrated circuits, specially designed ASICs (application-specific integrated circuits), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include implementations in one or more computer programs executable and / or interpretable on a programmable system including at least one programmable processor. It may be coupled to receive data and instructions from and transmit data and instructions to a storage system, at least one input device, and at least one output device, for a specific or general purpose.

[0185] These computer programs (also called programs, software, software applications, or code) contain machine language instructions for a programmable processor and may be implemented in high-level procedural and / or object-oriented programming languages ​​and / or in assembly / machine language. As used herein, the terms machine-readable medium and computer-readable medium refer to any computer program product, apparatus and / or device (e.g., magnetic disks, optical disks, memory, programmable logic devices (PLDs)) used to provide machine instructions and / or data to a programmable processor. This includes machine-readable medium that receives machine instructions as machine-readable signals. The term machine-readable signal refers to any signal used to provide machine instructions and / or data to a programmable processor.

[0186] To provide user interaction, the systems and technologies described herein may be implemented in a computer equipped with a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) and a keyboard and pointing device (e.g., a mouse or trackball) to which the user can provide input to the computer. Other types of devices may also be used to provide user interaction. For example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or haptic feedback). Input from the user may be received in any form, including acoustic, speech, or haptic input.

[0187] The systems and technologies described herein may be implemented within a computing system that includes backend components (e.g., data servers), middleware components (e.g., application servers), or frontend components (e.g., client computers with a graphical user interface or web browser through which a user can interact with an implementation of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the internet.

[0188] A computing system can include clients and servers. Clients and servers are generally geographically separated from each other and typically interact via a communication network. The client-server relationship arises from computer programs running on each computer that have a client-server relationship with one another.

[0189] Figures 19A and 19B are block diagrams of an exemplary system 1900 equipped with two pressure sensors and two acoustic sensors, and block diagrams of data collected from such a system. In system 1900, two users sleep on a single mattress. Acoustic and pressure sensors collect acoustic and pressure signals of physical phenomena created (generated) by the two sleepers. A snoring / breathing analysis engine 1916 can determine which user's actions contribute to the information in the data stream and can generate snoring / breathing parameters for the two users.

[0190] Bed 1902 is an airbed equipped with a left-side sleeping section and a right-side sleeping section to support, for example, a sleeper on the left side and a sleeper on the right side. When two users are sleeping, their bodies generate phenomena that affect the environment around them. For example, breathing generates acoustic and pressure waves within bed 1902, snoring generally generates stronger acoustic and some pressure waves in the air, heartbeat generates pressure waves, etc.

[0191] The right acoustic sensor 1904 and the left acoustic sensor 1905 can be positioned to sense the surrounding acoustics of two users. Sensors 1904 and 1905 can be configured or positioned such that the left acoustic sensor 1905 is more likely to receive a stronger signal from the left user, and the right acoustic sensor 1904 is more likely to receive a stronger signal from the right user. For example, sensors 1904 and 1905 can be embedded in the outer wall of the mattress. In some examples, sensors 1904 and 1905 may be directional, i.e., more sensitive in one direction than in another, and this directionality of sensors 1904 and 1905 can be positioned to target each user.

[0192] The signal modifier 1906 modifies the raw sensor data by modifying the raw sensor signals so that the raw sensor data is more useful to the system 1900. For example, the signal modifier 1906 may amplify the sound in a frequency band of interest through the use of amplifiers and filters.

[0193] The digitizer 1908 can convert a tuned analog acoustic signal into a digital acoustic signal stream. For example, the digitizer 1908 can convert the analog acoustic signal into a digital representation (e.g., a numerical value) so that it can be stored or processed by digital logic circuits such as a microprocessor or computer. For example, the digitizer can receive an analog acoustic stream having a specific waveform shape and generate a stream of digital values ​​describing the waveform according to a predetermined conversion algorithm. In some implementations, this digital stream is a two's complement binary value (binary number) proportional to the value of the input waveform at a specific sample rate.

[0194] Snoring and breathing are further monitored by the right-side pressure sensor 1911 and the left-side pressure sensor 1910, which sense the instantaneous air pressure on the right and left sides of the bed 1902 and convert it into an electrical signal. For clarity, two separate physical sensors are shown, but it will be understood that other configurations are possible. For example, as mentioned above, a single pressure transducer may be connected to two or more air bags, and one or more switches may cause the single transducer to function as a separate pressure signal for each air bag.

[0195] The signal modifier 1912 modifies the raw sensor signals to make the raw sensor data more useful to the system 1900. For example, the signal modifier 1912 may amplify the pressure signal in a frequency band of interest through the use of amplifiers and filters.

[0196] The digitizer 1914 can convert a regulated analog pressure signal into a digital pressure signal stream. For example, the digitizer 1914 can convert the analog pressure signal into a digital representation (e.g., a numerical value) so that it can be stored or processed by a digital logic circuit. For example, the digitizer 1914 can receive one of the analog pressure streams having a particular waveform shape and generate a stream of digital values ​​describing the waveform according to a predetermined conversion algorithm. In some implementations, this digital stream is a two's complement binary value (in base 1) proportional to the value of the input waveform at a particular sample rate.

[0197] The processor unit simultaneously receives and buffers incoming acoustic data streams and pressure data streams, and provides the buffered data epochs to a snoring / breathing analysis algorithm 1916 that runs on the processor unit and is capable of generating snoring / breathing parameters for two users on the bed. For example, the snoring / breathing analysis engine 1916 may run on the processor unit (either locally or remotely to the bed) and determine the snoring / breathing parameters 1918 for the user on the right side of the bed and the snoring / breathing parameters 1920 for the user on the left side of the bed. The snoring / breathing parameters 1918 and 1920 may include parameters representing the state and severity of snoring, the presence of snoring (e.g., a Boolean value of 1 or 0), the intensity of snoring in sound pressure level (SPL) (e.g., 60 dB), the severity level of snoring as used in clinical grading (e.g., between mild and loud), the presence of apnea (e.g., 0 or 1), and a severity scale (e.g., a numerical value between 0 and 5). These parameters can be appropriate digital circuit representations, including standard data formats such as integers or floating-point numbers, and can be stored in various different formats such as Extended Markup Language (XML). In this example, the snoring / breathing analysis engine 1916 generates a set of snoring / breathing parameters 1918 and 1920 for each epoch in which acoustic and pressure signals are received, with snoring / breathing parameter 1918 for the user on the right and snoring / breathing parameter 1920 for the user on the left.

[0198] The snoring / breathing analysis engine 1916 can process acoustic and pressure signals to separate the acoustics of the left-side user and the right-side sleeper through sound source separation processes such as independent component analysis (ICA). Each user's snoring / breathing parameters can be found by fusing the snoring / breathing information present in each sleeper's acoustic and pressure signals. Dual modality (i.e., acoustic and pressure) fusing can be achieved using a variety of data fusion techniques, including but not limited to the following examples.

[0199] In one example, the intermodality signal fusion process combines pre-processed and scaled versions of each user's acoustic and pressure data epochs into a single joint signal. This may result in the creation of two joint signals representing two users. In this case, the process may include each of the two joint signals to identify each individual's snoring and breathing state and parameters.

[0200] In one example, the intermodality signal fusion process generates a set of features and / or parameters associated with each user's acoustic and pressure epoch set. Based on all or some of the combinations of features and / or parameters from the acoustic and pressure epochs, the process can use a supervised machine learning classifier to determine the snoring / breathing parameters 1918 and 1920.

[0201] For example, the intermodality signal fusion process may include detecting snoring / breathing parameters in one modality and verifying and / or activating those snoring / breathing parameters in a second modality. For instance, the snoring / breathing analysis engine 1916 may detect snoring events based on processing separate acoustic signals for each user, as described above. Then, to ensure that more accurate and / or efficient processing is utilized, the parameters of the acoustic signal may be compared with parameters obtained from the pressure signal.

[0202] In one example, the intermodality fusion process may involve detecting snoring events based on processing the pressure signal by filtering it. For instance, a high-pass filter with a cutoff frequency of 60 Hz may be used to capture the snoring-induced component in the pressure signal. Then, for more accurate and / or efficient processing, the process may compare parameters of the pressure-based respiratory signal (e.g., periodicity, fundamental frequency) with those obtained by the sleeper's acoustics.

[0203] For example, the intermodality fusion process may include the snoring / breathing analysis engine 1918 being activated or deactivated based on the presence of a user in the bed 1902 and the user's sleep state. For instance, the snoring / breathing analysis engine 1918 may be activated (activated) when at least one user is in the bed and / or when at least one user has fallen asleep, as determined by analysis of acoustic and / or pressure signals.

[0204] Figure 19B shows the data that can be collected by acoustic sensors 1904 and 1905 and pressure sensors 1911 and 1910.

[0205] In panel A, the pressure on the left is shown in blue, and the estimated respiratory signal is shown in black.

[0206] In panel B, the pressure on the right is shown in cyan, and the estimated respiratory signal is shown in black.

[0207] In panel C, the recording of the acoustic sensor signal on the left is shown in red.

[0208] In panel D, the recording of the acoustic sensor signal on the right is shown in magenta.

[0209] In panel E, the sound of person 1 snoring, recovered from the left and right acoustic signals, is shown in black. This data was created using a blind source separation process, e.g., ICA.

[0210] In panel F, the sounds of person 2 snoring, recovered from the left and right acoustic signals, are shown in black. This data was created using a blind source separation process, e.g., ICA.

[0211] Figures 20A and 20B are block diagrams of an exemplary system 2000 with two pressure sensors and one acoustic sensor, and block diagrams of data collected from such a system. Unlike system 1900, system 2000 includes a single acoustic sensor 2004. Acoustic sensor 2004 is configured to record both the left user and the right user. System 2000 also includes pressure sensors 2010 and 2011, signal modifiers 2006 and 2012, digitizers 2008 and 2014, snoring / respiratory analysis engine 2016, and snoring / respiratory parameters 2018 and 2020.

[0212] The acoustic sensor 2004 may be placed, for example, within the bed 2002 (e.g., within the headboard, within the mattress side wall). In some cases, the acoustic sensor 2004 may be placed equidistant from two users. This may result in capturing the acoustics of two users with equal sensitivity. The system 2000, comprising one acoustic sensor 2004 and two pressure sensors 2010 and 2011, can be used to reliably monitor snoring / breathing status for the duration that at least one user is on a designated side of the bed, or for the duration that at least one sleeper is snoring.

[0213] In one example, the intermodality fusion process could combine pre-processed and scaled versions of each user's pressure data epoch with those of common acoustic data. This could result in two joint signals representing two users. The process could then use these two joint signals to identify the user who is snoring and their snoring / breathing parameters 2018 or 2020.

[0214] For example, an intermodality fusion process may generate a set of features and / or parameters associated with a common set of acoustic and individual pressure epochs. Based on all or some of the combinations of features and / or parameters from the acoustic and pressure epochs, the process may use a supervised machine learning classifier to determine the snoring / breathing parameters 2018 and 2020.

[0215] For example, an intermodality signal fusion process may include detecting snoring / breathing parameters in one modality and verifying and / or activating those snoring / breathing parameters in a second modality. This process may include generating a set of features and / or parameters associated with each user's acoustic and pressure epoch set, based on all or some of the combinations of features and / or parameters from a common acoustic signal. If the common acoustic signal indicates a snoring event by at least one sleeper or exactly one sleeper, the parameters of that signal may be compared to the parameters of each pressure signal to identify the person actively snoring.

[0216] For example, the snoring / breathing analysis engine 2016 may detect snoring events based on processing a common acoustic signal. Then, to ensure that more accurate and / or efficient processing is utilized, the parameters of the acoustic signal may be compared with parameters obtained from the pressure signal.

[0217] For example, the intermodality fusion process may include the snoring / breathing analysis engine 2016 being activated or deactivated based on the presence of a user in the bed 2002 and the user's sleep state. For instance, the snoring / breathing analysis engine 2016 may be activated (started) when at least one user is in the bed and / or when at least one user has fallen asleep, as determined by analysis of acoustic and / or pressure signals.

[0218] Figure 20B shows the data that can be collected by the acoustic sensor 2004 and the pressure sensors 2011 and 2010.

[0219] In panel A, the pressure on the left is shown in blue, and the estimated respiratory signal is shown in black.

[0220] In panel B, the pressure on the right is shown in cyan, and the estimated respiratory signal is shown in black.

[0221] In panel C, acoustic recordings from a common sensor are shown in red.

[0222] Figure 21 is a block diagram of an exemplary system 2100, which includes one pressure sensor and two acoustic sensors. Unlike system 1900, system 2100 uses a single pressure sensor 2110. The pressure sensor 2110 is configured to record both the left-side and right-side users. System 2100 also includes acoustic sensors 2104 and 2105, signal modifiers 2016 and 2112, digitizers 2108 and 2114, a snoring / breathing analysis engine 2116, and snoring / breathing parameters 2118 and 2120.

[0223] Figure 22 is a flowchart of an exemplary process 2200 for operating a home automation device. For example, process 2200 may be used in any of systems 1900, 2000, or 2100.

[0224] Presence is determined (2202). For example, process 2200 may include determining whether one or two users are in the bed using pressure and / or acoustic readings. If presence is not detected, process 2200 may wait until presence is detected. If presence is detected, process 2200 may proceed.

[0225] Sleep is determined (2204). For example, process 2200 may include determining whether one or both of the two users on the bed are asleep, using pressure and / or acoustic readings. If sleep is not detected, process 2200 may wait until sleep is detected. If sleep is detected, process 2200 may proceed.

[0226] The snoring / breathing parameters for the left side are determined (2208), and the snoring / breathing parameters for the right side are determined (2206). For example, process 2200 may include determining the snoring / breathing parameters for the left-side user using pressure and / or acoustic readings. Substantially simultaneously, process 2200 may include determining the snoring / breathing parameters for the right-side user using the same pressure and / or acoustic readings.

[0227] As can be understood, a portion of the acoustic energy in the environments of the left-hand user and the right-hand user is generated by the left-hand user, and at least a portion of the acoustic energy in the environments of the left-hand user and the right-hand user is created by the right-hand user. Due to this crosstalk, the signals of the right-hand user and the left-hand user may be unraveled as part of creating the left-hand snoring / breathing parameter (2208) and the right-hand snoring / breathing parameter (2206).

[0228] As shown here, parameter determinations (2206 and 2208) are performed after and / or in response to the sleep determination (2204). This configuration may be advantageous because it can reduce the use of the system. Parameter determinations (2206 and 2208) may only be performed when a person may be snoring, i.e., when they are asleep.

[0229] However, other configurations are possible. For example, the parameter determination (2206 and 2208) may be performed after the presence of a user in bed and / or in response to the presence of a user in bed, instead of the sleep determination (2204). This may be advantageous because process 200 can be utilized by a system that operates or reacts in response to a user being in bed, regardless of whether the user is asleep or not. For example, a system to improve home security may be configured to verify that all external doors are locked when a user lies down in bed and to warn the user about any unlocked doors. This notification may be configured to warn the user when they are in bed but before they fall asleep. In another form, process 2200 may perform the parameter determination (2206 and 2208) without sleep or presence detection (2202 and 2204). This may be beneficial in situations where presence and snoring detection is impossible or undesirable. For example, a sleep clinic that treats sleep disorders (sometimes called sleepwalking) could be advantageously configured to monitor snoring even when the user is not in bed due to the sleep disorder.

[0230] A device is activated if snoring / breathing parameters match a certain rule (2212). For example, one user may set up one or more home automation rules that should be activated (triggered) when a specific sleep or snoring pattern of one user is detected. In one example, a user may create a rule that activates a heater and humidifier if a specific set of snoring / breathing parameters indicates that the user, who is an allergy sufferer, is experiencing slight difficulty breathing. In another example, a user may set up a rule that raises the head of the bed in an attempt to reduce or eliminate snoring if snoring is detected. In response to a rule that satisfies its execution conditions, the controller may send a command to a controllable device. The controllable device receives a command to drive itself. For example, an HVAC system may receive a command to increase the temperature and humidity, or a bed base may receive a command to raise the head of the bed.

[0231] Figure 23 is a pipeline diagram of an example of a pipeline (logic circuit chain) 2300 that may be used to collect acoustic and pressure readings for home automation. The pipeline 2300 may be used by a bed system that includes the function of sending control signals to a home automation device in response to the detection of snoring by a user on the bed. For example, pipeline 2300 may be included in an airbed controller that also controls the firmness and height of the bed. In some examples, pipeline 2300 may be used by other data processing systems. For example, the acoustic sensor 2302 may be integrated into different elements of a home automation system that communicates with the bed system. For example, pipeline 2300 may be used with systems 2300, 2000, 2100, signal modifiers 2306, 2312, 2006, 2012, 2106, and 2112; digitizers 2308, 2314, 2008, 2014, 2108, and 2114; snoring / respiratory analysis engines 2316, 2016, and 2116, etc.

[0232] The acoustic sensor 2302 may include hardware and software configured to generate data streams in response to acoustic energy in the environment. The acoustic sensor 2302 may include, for example, one or more microphones embedded in a bed frame or mattress on a bed. The acoustic sensor 2302 may include, for example, multiple sensors placed throughout a building such as a house or hospital. In some cases, the acoustic sensor 2302 may include different types of sensors from different sources. For example, a sensor embedded in a bed and a sensor in a telephone may work together to generate one or more data streams from their individual sensings.

[0233] The acoustic sensor 2302 can generate an analog acoustic stream 2304. Some acoustic sensors 2302 generate an analog signal, which is an analog electrical signal proportional to the acoustic energy received by the sensor 2302. For example, if the acoustic energy is a pressure wave having a specific shape, the sensor 2302 can generate an analog acoustic stream 2304 having radio waves of the same specific shape.

[0234] The digitizer 2306 can receive an analog acoustic stream 2304 and generate a digital acoustic stream 2308. For example, the digitizer 2306 can receive an analog acoustic stream 2304 having a specific waveform shape and generate a stream of digital values ​​describing the waveform according to a predetermined conversion algorithm. In some implementations, this digital stream is a two's complement binary value (in base 1) proportional to the value of the input waveform at a specific sample rate.

[0235] In some implementations, the sensor 2302 does not generate an analog acoustic stream 2304, but instead generates a digital acoustic stream 2308. In some cases, two or more acoustic streams are used, regardless of whether they are digital or analog. For clarity, the following description is made with reference to pipeline 2300 using a single sensor 2302 that generates a single analog acoustic stream 2304, but other configurations are also possible.

[0236] The framer 2310 generates a digital acoustic frame 2312 from the digital acoustic stream 2308. For example, if the digital acoustic stream 2308 is a stream of binary digits, the framer 2310 may generate a digital acoustic frame 2312 that contains all of the binary digits within a fixed time window.

[0237] In some implementations, digital audio frames 2312 can overlap. For example, each frame may be 100ms long and may overlap with the previous digital audio frame by 50ms and with the next digital audio frame by 50ms. In another example, each frame may be 200ms long and may overlap with two adjacent digital audio frames by 10ms each. In yet another example, each frame may be 20s long and may overlap with two adjacent digital audio frames by 1s each.

[0238] The pipeline 2300 may also include a pressure sensor 2314. For example, the pressure sensor 2314 may be contained within a bed, such as an air bed, and may include hardware and software configured to generate a data stream in response to pressure applied to the bed by one or more users on the bed. The pressure sensor 2314 may include, for example, a transducer or flexible membrane fluidly coupled to an air bag by a hose. In some cases, the pressure sensor 2314 may be separable from the bed, for example, in the form of a pad, strip, pack, or sheet that can be placed on or under the mattress of the bed.

[0239] The pressure sensor 2314 can generate an analog pressure stream 2316. Some pressure sensors 2314 generate an analog signal, which is an analog electrical signal proportional to the pressure received by the sensor 2314. For example, if the pressure is a pressure wave having a specific shape, the sensor 2314 can generate an analog pressure stream 2316 having radio waves of the same specific shape.

[0240] The digitizer 2318 can receive an analog pressure stream 2316 and generate a digital pressure stream 2320. For example, the digitizer 2318 can receive an analog pressure stream 2316 having a specific waveform shape and generate a stream of digital values ​​describing the waveform according to a predetermined conversion algorithm. In some implementations, this digital stream is a two's complement binary value (in base 1) proportional to the value of the input waveform at a specific sample rate. In some cases, the digitizers 2306 and 2318 can use the same sampling rate. In some cases, the digitizers 2306 and 2318 can use different sampling rates.

[0241] In some implementations, sensor 2314 does not generate an analog pressure stream 2316, but instead generates a digital pressure stream 2320. In some cases, two or more pressure streams are used, regardless of whether they are digital or analog. For clarity, the following description is made with reference to pipeline 2300 using a single sensor 2314 that generates a single analog pressure stream 2316, but other configurations are also possible.

[0242] Framer 2322 generates a digital pressure frame 2324 from the digital pressure stream 2320. For example, if the digital pressure stream 2320 is a stream of binary digits, framer 2322 may generate a digital pressure frame 2324 that contains all of the binary digits within a fixed time window.

[0243] In some implementations, the digital pressure frames 2324 may overlap. For example, each frame may be 100ms long and may overlap with the previous digital pressure frame by 50ms and with the next digital pressure frame by 50ms. In another example, each frame may be 200ms long and may overlap with two adjacent digital pressure frames by 10ms each. In yet another example, each frame may be 30 seconds long and may overlap with the previous and next digital pressure frames by 1 second each.

[0244] The digital acoustic frame 2312 and the digital pressure frame 2324 may be used by the encryption / compression engine 2332 to prepare the digital acoustic frame 2312 and the digital pressure frame 2324 for storage. The encryption / compression engine 2332 may create encrypted / compressed reads 2334 which contain securely encrypted and compressed data. When decrypted and compressed, it generates the digital acoustic frame 2312 and the digital pressure frame 2324. The encryption / compression engine 2332 may send the encrypted / compressed reads 2334 to off-site or local storage 2336, such as cloud storage.

[0245] The snoring analyzer 2326 may also use digital acoustic frames 2312 and digital pressure frames 2324 to determine the snoring state of the user in bed. As shown below, for example, one or more machine learning processes may be used, and based on the determination of the snoring state, the snoring analyzer 2326 may generate a corresponding control signal 2328. The controller array 2330 may receive the control signal and act on controllable devices in accordance with the control signal to change the user's environment.

[0246] The snoring analyzer 2326 may report snoring, respiration, or other biometric parameters to local or one or more cloud-based services. For example, historical user information may be stored in local computer memory (e.g., computer memory 512) for long-term storage, aggregation, analysis, etc. Additionally or alternatively, historical user information may be stored in computer memory away from the snoring analyzer 2030 for long-term storage, aggregation, analysis, etc. (e.g., pressure-based sleep data 1312).

[0247] The snoring analyzer 2326 can make decisions about the snoring state using one or a combination of calculations. For example, within each frame, features corresponding to the temporal and spectral characteristics of the acoustic readings may be generated. Examples of such features are: Functions of minimum, maximum, mean, median, standard deviation, and peak amplitude, width, and position of the audio signal within an epoch. The minimum, maximum, mean, median, and standard deviation of the audio signal envelope within an epoch, as well as functions of the amplitude, width, and position of the peaks. Functions of minimum, maximum, mean, median, standard deviation, and peak amplitude, width, and position of the spectrum of an audio signal within an epoch. The minimum, maximum, mean, median, standard deviation, and functions of the amplitude, width, and position of the peaks of the spectrum of the audio signal envelope within an epoch. The acoustic snoring index is calculated from the spectrum of the audio signal within an epoch as a ratiometric measurement of different spectral subbands. Mel frequency coefficient from the spectrum of the audio signal within the epoch, This includes, but is not limited to, the following:

[0248] For example, within each frame, features corresponding to the temporal and spectral characteristics of the pressure readings may be generated. Examples of such features include: A function of the number (rate) of breaths measured from pressure fluctuations. A function of respiratory amplitude measured from pressure fluctuations, A function of respiration duration measured from pressure fluctuations, Functions of minimum, maximum, mean, median, standard deviation, and peak amplitude, width, and position of the pressure signal within an epoch. Functions of minimum, maximum, mean, median, standard deviation, and peak amplitude, width, and position of the pressure signal spectrum within an epoch. The pressure snoring index is calculated from the spectrum of the pressure signal within an epoch as a ratiometric measurement of different spectral subbands. This includes, but is not limited to, the following:

[0249] Figures 24A and 24B are flowcharts of exemplary processes for determining snoring / respiratory parameters. For example, processes 2400 and 2450 may be used in any of systems 1900, 2000, or 2100. For clarity, the following description uses system 1900, but one or more other systems may be used to perform the same or similar processes.

[0250] In process 2400, a machine learning classifier is generated by being trained on population-level training data to classify data streams. This training data can be collected, tagged by a guided or unguided tagging process, and presented to the machine learning trainer. The trainer can use the tagged training data to create a classifier that distinguishes between various snoring states.

[0251] Data streams are received (2402). For example, digitizers 1908 and 1914 may receive data from signal modifiers 1906 and 1912. Snoring / respiratory analyzer 1916 may receive digitized data from digitizers 1908 and 1914, which includes sensing information performed by acoustic sensors 1905, 1904, 1910, and 1911. Thus, in this example, four data streams are received. However, other configurations may use a different number of data streams.

[0252] The data streams are provided to the snoring classifiers (2404). For example, upon reception, the snoring / respiratory analyzer 1916 may provide four data streams to one or more snoring classifiers, each configured to receive the four data streams and provide a classification of whether the data streams contain snoring or not. These classifications may take any technically appropriate form, including the form of confidence values ​​(e.g., on a scale of 0 to 1) that measure the confidence that the classifier test produced a correct classification.

[0253] Classifications can be aggregated (aggregated). One exemplary aggregation is to select the classification with the highest confidence score. Another exemplary aggregation is to collect votes from confidence levels above a threshold and select the classification that receives the most votes (snoring / no snoring). Other aggregations are also possible.

[0254] If snoring is detected (2406), the data stream is provided to the left / right classifier (2408). The left / right classifier may operate on the same data stream and may generate one of two (left or right) or three (left, right, or both) possible classifications, along with a confidence value.

[0255] In this configuration, the left / right classifiers are configured to classify only as left or right. These classifiers in this example are not configured to generate a classification indicating no snoring. In other words, these left / right classifiers operate on the premise that snoring is present.

[0256] This configuration may offer one or more potential advantages by using left / right snoring-side classifiers that operate on the premise of snoring. For example, if the classifier only needs to classify a given phenomenon (i.e., snoring-side) and does not need to determine whether the phenomenon exists or not (i.e., presence or absence of snoring), then many machine learning processes (e.g., feature learning) used to generate the classifier may produce a more accurate classifier. Thus, process 2400 may have particular value when using these types of machine learning techniques. When a classifier is designed to determine whether snoring exists or not and classify the snoring in a single classification process, a different process, process 2450, may be used.

[0257] Snoring / breathing parameters are generated (2410). For example, the snoring / breathing analysis engine 1916 may generate snoring parameters within the snoring / breathing parameters to reflect the results of the classification (2404 and 2408). The snoring / breathing analysis engine 1916 may also generate other parameters to record other perceived respiratory features (e.g., respiratory rate, variability, etc.). The snoring / breathing analysis engine 1916 may use classifiers on these respiratory parameters, which may be constructed from machine learning techniques, non-machine learning techniques, or a mixture of both.

[0258] In process 2450, a machine language classifier is generated by being trained on population-level training data to classify data streams. This training data can be collected, tagged by a guided or unguided tagging process, and presented to a machine learning trainer. The trainer can use the tagged training data to create a classifier that distinguishes between various snoring states.

[0259] A data stream is received (2452), and if snoring is detected (2456), the results for left and right snoring are collected (2458). For example, digitizers 1908 and 1914 may receive data from signal modifiers 1906 and 1912. The snoring / respiratory analyzer 1916 may receive digitized data from digitizers 1908 and 1914, which includes sensing information performed by acoustic sensors 1905, 1904, 1910, and 1911. Thus, in this example, four data streams are received. However, other configurations may use a different number of data streams.

[0260] The data stream is provided to the left / right / no-snoring classifier (2454). For example, upon reception, the snoring / respiratory analyzer 1916 may provide four data streams to one or more left / right / no-snoring classifiers configured to receive the four data streams and provide a classification of snoring status in a single operation. This classification includes whether or not snoring is present (e.g., left-sided snoring, right-sided snoring, bilateral snoring, no snoring).

[0261] This configuration may offer one or more potential advantages by using a left / right / no-snoring classifier that operates without assuming snoring. For example, if all unknown classifications (presence or absence of snoring, snoring side) are included in a single training, many machine learning processes (e.g., deep neural networks) for generating the classifier may produce a more accurate classifier. Therefore, when using these types of machine learning techniques, process 2450 may have certain value. Different processes may be used when the classifier is designed to determine only whether or not snoring is present in a single process, and / or only which side the snoring is on in a single process.

[0262] Snoring / breathing parameters are generated (2460). For example, the snoring / breathing analysis engine 1916 may generate snoring parameters within the snoring / breathing parameters to reflect the results of the classification (2454-2458) and the characteristics of the snoring (e.g., snoring intensity). The snoring / breathing analysis engine 1916 may also generate other parameters to record other perceived respiratory characteristics (e.g., respiratory rate, variability, etc.). The snoring / breathing analysis engine 1916 may use classifiers on these respiratory parameters, which may be constructed from machine learning techniques, non-machine learning techniques, or a mixture of both.

[0263] Figure 25 is a swimlane diagram of an exemplary process 2500 for training and using a machine learning classifier to determine and classify snoring events in a bed. For clarity, process 2500 is described with reference to a specific set of components. However, one or more other systems may also be used to perform the same or similar processes.

[0264] In process 2500, the bed system uses readings from the pressure / acoustic sensor 2502 to learn how the user affects the bed pressure and what sounds the user produces when they are snoring or not. The bed system can use these readings as signals for a determination engine that classifies the user's snoring state into one of several possible snoring states. In another example, the snoring state may include two states (e.g., snoring or not), or any of the following states: snoring or not, snoring only on the left side, only on the right side, or on both sides, or a number of other snoring states that describe the user's snoring in more detail (e.g., louder snoring on the left side and milder snoring on the right side).

[0265] During operation, the bed may determine the user's snoring status and operate accordingly. For example, the user may configure the bed system to change the pressure to minimize their snoring when they are snoring. The bed may operate to determine the snoring status iteratively or continuously based on a series of live readings from the pressure / acoustic sensor 2502. When the snoring status changes, for example, from "no snoring" to "snoring," the bed system may instruct the pump to change the pressure of the mattress beneath the user.

[0266] A pressure / acoustic sensor 2502 senses pressure (2512). For example, the pressure sensor may create a live stream of pressure readings that reflect the internal pressure of the airbag in the bed system. This live stream of pressure readings may be provided to the bed controller 2504 substantially on a regular basis in the form of analog or digital information. This may reflect the pressure inside the airbag caused by the user (or other object) on the bed system, or when the bed is empty.

[0267] Simultaneously, the acoustic sensor can create a live stream of acoustic readings that reflect the acoustic energy in the environment surrounding the user of the bed system. This live stream of acoustic readings can be provided to the bed controller 2504 in the form of analog or digital information on a substantially constant basis. This allows for the reflection of the acoustic conditions around the user, resulting from sounds created by the user, such as snoring or speech.

[0268] The bed controller 2504 receives pressure / acoustic readings (2514). For example, the bed controller 2504 may store the pressure / acoustic readings in a computer memory structure such as a rolling buffer that makes the most recent N readings available to the bed controller. The bed controller 2504 may aggregate these pressure / acoustic readings, subsample them, or store them all individually.

[0269] The bed controller 2504 transmits pressure / acoustic readings (2516), and the cloud reporting service 2506 receives these pressure / acoustic readings (2518). For example, the bed controller 2504 may transmit all pressure / acoustic readings to the cloud reporting service 2506, or it may determine that only some pressure / acoustic readings (but not others) should be transmitted to the cloud reporting service 2506. The cloud reporting service 2506 is configured to receive pressure / acoustic readings and, in some cases, other types of data. The pressure / acoustic readings transmitted to the cloud reporting service 2506 may not be modified by the bed controller 2504, may be aggregated (e.g., average, maximum, and minimum), or may be modified by the bed controller 2504 in other ways.

[0270] Classifier factory 2508 generates a classifier from the pressure / acoustic readings 2520. For example, classifier factory 2508 may perform feature learning-based machine learning to generate a classifier. Classifier factory 2508 may train the classifier by first obtaining a large set of pre-classified reading variation patterns. For example, one bed or many beds may report reading data to cloud reporting service 2506. This reading data may be tagged, recorded, and stored for analysis in creating a pressure classifier for use by bed controller 2504 and / or other bed controllers.

[0271] Classifier factory 2508 may generate features from the readings. For example, a stream of pressure signals and a stream of acoustic signals may be divided into multiple buffers of, for example, 1 second, 2.125 seconds, 3 seconds, or 25 seconds, and features in the time or frequency domain may be generated. These features may be direct measurements of pressure / acoustic within these buffers. For example, such features may include maximum, minimum, or random pressure / acoustic values. These features may be derived from the readings within these buffers. For example, such features may include an average reading, a standard deviation, or a gradient value indicating an increase or decrease over time within the buffer. The values of the feature vector may be in binary or numerical format. For each buffer, multiple values may be stored in a predetermined order to create a vector composed of a series of fields. Each vector has the same series of fields and the data within those fields. Some other features may be calculated from the transformed domain representation of the pressure and acoustic signals, such as the coefficients of a Fourier transform or a wavelet transform.

[0272] As another example, the classifier factory may identify instances within the readings that match a certain pattern or rule for a certain pattern. In one example, the repeating pattern may be defined as a sine wave shape or a sawtooth shape of a pressure or acoustic stream, which includes significant increases or rapid fluctuations. Such a pattern can be identified, and corresponding composite information regarding the pattern in time or frequency (e.g., timestamp, duration, maximum envelope amplitude, spectral peak) can be synthesized from the pressure and acoustic signals and / or other external information (e.g., real-time clock).

[0273] The classifier factory 2508 can combine or reduce features. For example, the extracted features can be combined using principal component analysis. For principal component analysis of the features, the classifier factory 2508 can determine a subset of all the features that distinguish (discriminate) the user's snoring state. That is, the classifier factory 2508 can classify the features into features useful for determining the snoring state and features of low usefulness, and the more useful features can be retained. This process can be performed on a trial-and-error basis, and random combinations of features may be tested. This process may be performed using one or more systematic processes. For example, linear discriminant analysis or generalized discriminant analysis may be used.

[0274] In some cases, an appropriate subset of features can be selected from the set of all available features. If multiple classifiers are created, the selection can be made once for each classifier. Alternatively, the selection can be made once for multiple or all classifiers if multiple classifiers are created.

[0275] For example, random (or pseudo-random) numbers may be generated, and the features of those numbers may be removed. In some cases, multiple features may be aggregated into a single aggregate feature. For example, if multiple repeating patterns are identified in pressure or acoustic readings, those repeating patterns and / or the composite data associated with those repeating patterns may be aggregated. For example, the durations of all snoring patterns may be aggregated into the mean, standard deviation, minimum, and / or maximum duration.

[0276] The classifier factory 2508 can process features. For example, the remaining features can be processed to rationalize their values, with each feature being processed with a weight corresponding to how discriminative it is. If a feature is found to be very discriminative and very useful for classifying states, it may be given a larger weight than the others. If a second feature is found to be less discriminative than the others, it may be given a smaller weight.

[0277] When mapped to kernel space, features can be standardized, the data points can be concentrated to a predetermined mean, and the features can be scaled to have a unit standard deviation. This allows all features to have, for example, a mean of 0 and a standard deviation of 1. The extracted features are then converted to a vector form using the same vector form described above.

[0278] In some cases, the remaining features can be processed by applying a kernel function to map the input data into a kernel space. The kernel space allows for the clustering of a higher-dimensional space (e.g., a vector space into which vectors of feature data are input) so that different clusters can represent different states. The kernel function can be any suitable form, including linear, quadratic, polynomial, radial basis, multilayer perceptron, or custom.

[0279] The classifier factory 2508 can train classifiers. For example, a pattern recognition algorithm may use a vector of extracted features and their corresponding existence state labels as a dataset to train a classifier by which new pressure readings can be classified. In some cases, this may involve storing the classifier along with the training data for later use.

[0280] The classifier factory 2508 may transmit classifiers (2522), and the bed controller 2504 may receive classifiers (2524). For example, one or more classifiers created by the classifier factory 2508 may be transmitted to the bed controller 2504 and / or other bed controllers. In some cases, classifiers may be transmitted over a data network. In some cases, classifiers may be transmitted (provided) on non-temporary computer-readable media such as a compact disc (CD), a universal serial bus (USB) drive, or other device. Classifiers may be loaded (installed) onto the bed controller 2504 and / or other bed controllers as part of a software installation, as part of a software update, or as part of another process. In some cases, the classifier factory 2508 may send a message to the bed controller 2504 and / or other bed controllers, which may contain data specifying one or more classifiers that classify a bed into one of several snoring states using a stream of pressure readings and / or a stream of acoustic readings. In some configurations, the classifier factory 2508 may transmit classifiers at once, either in a single message or in a series of messages that are close together in time. In some configurations, the classifier factory 2508 may transmit classifiers that are separated in time. For example, the classifier factory 2508 may generate and transmit classifiers. Subsequently, as more pressure sensor data becomes available, the classifier factory 2508 may generate updated or new classifiers that are different from those already created.

[0281] A classifier can be defined by one or more data structures. For example, the classifier factory 2508 may record a classifier in an executable or interpretable file, such as a software library, an executable file, or an object file. A classifier can be stored, used, or transmitted as a structured data object, such as an Extensible Markup Language (XML) document or a JavaScript Object Notation (JSON) object. In some examples, a classifier can be created in a binary or script format that the bed controller 2504 can execute (e.g., run or interpret). In some examples, a classifier can be created in a format that is not directly executable, but in a format that allows the bed controller 2504 to construct a classifier according to that data.

[0282] The bed controller 2504 may also classify snoring using streams of pressure readings and acoustic readings (2526). For example, the bed controller 2504 may run one or more classifiers using data from streams of pressure readings and acoustic readings. The classifiers may classify the data into one of several states (e.g., no snoring, light snoring, etc.). For example, the classifier may convert the data streams into the aforementioned vector form. The classifier then examines the vectors to mathematically determine whether they resemble training data labeled as one state or training data labeled as another state. Once this similarity is calculated, the categorizer may return a response indicating that state.

[0283] The snoring analyzer uses one or more machine learning classifiers to categorize frames of pressure and / or acoustic readings into snoring intensity levels. In one example, the classifier classifies epochs into two classes: no snoring, snoring. In another example, the classifier classifies epochs into three classes: no snoring, intermittent snoring, continuous snoring. In yet another example, the classifier classifies epochs into four classes: no snoring, light snoring, moderate snoring, loud snoring. In yet another example, the classifier classifies epochs into five classes: no snoring, light snoring, moderate snoring, loud snoring. In yet another example, the classifier classifies epochs into six classes: no snoring, light snoring, moderate snoring, moderate to loud snoring, loud snoring. Such classifications follow clinical-grade snoring classifications.

[0284] The bed controller 2504 may use two or more classifiers. That is, the bed controller 2504 may have access to multiple classifiers, each functioning differently and / or using different training data to establish a classification. In such cases, the classifier's decisions may be treated as votes, and the vote count may be used to determine the presence or absence of snoring. If only one classifier is used, that classifier's vote is the only vote, and that vote is used to detect the snoring state. If there are multiple classifiers, different classifiers may produce conflicting votes, and the bed controller may select one snoring state that has received a vote. Classifiers may be created using a variety of machine learning and non-machine learning techniques. For example, in addition to feature learning described here, classifiers may be created using other machine learning techniques. An example of creating a classifier using deep learning will be described later. Other classifiers may be created using heuristics or pattern matching for generating classifications.

[0285] Various vote counting methods are possible. In some cases, the bed controller 2504 can count votes for each snoring state, and the snoring state with the most votes is determined to be the snoring state. In some cases, the bed controller 2504 can use other vote counting methods. For example, votes from different classifiers can be weighted based on the historical accuracy of the classifiers. In such a method, classifiers that have historically shown to be more accurate may be given a greater weight, while classifiers with lower historical accuracy may be given a smaller weight. This accuracy can be tracked at the population level or at the specific user level.

[0286] In some cases, votes may be cast by systems other than machine learning systems, and these votes may be incorporated into the total number of votes, influencing the outcome of the voting decision. For example, a non-machine learning pressure classification algorithm may cast a vote, for instance, based on a comparison with a threshold.

[0287] In some cases, the system may have different operating modes and may count votes differently depending on the mode. For example, different voting strategies may be used when the bed is being adjusted, or when the adjustable base is moving or part of the bed is being raised. In some modes, some classifiers may be given larger or smaller weights, or no weight at all, compared to some other modes. This can be useful, for example, when a classifier has been shown to be accurate in one mode (e.g., the bed is flat) for another mode (e.g., the head of the bed is raised by the base).

[0288] In some cases, the bed controller 2504 may ensure that the user is in and / or asleep before determining whether snoring is occurring. For example, using one or both of the pressure and / or acoustic readings, the bed controller may first determine whether the user is in the bed or whether the bed is empty. If it is determined that the user is in the bed, the bed controller 2504 may determine whether the user is asleep in the bed. Depending on the configuration, once the presence and sleep of the user are confirmed, the bed controller 2504 may determine whether snoring is occurring (2526).

[0289] The bed controller 2504 selects a device action (2528). For example, in response to a determination that the user is not snoring, or in response to a determination that the user is snoring, the bed controller 2504 may select a device action to be performed. For example, a set of rules stored in computer-readable storage, either locally or on a remote machine, may identify the action requested by the user or another system based on the snoring state. For example, the user may record (document) via a graphical user interface that they want the while-noise machine to operate when they are snoring. That is, only when they are snoring, but so that white noise is used to mask the snoring so as not to disturb their partner.

[0290] Based on the rule set and snoring detection, the bed controller 2504 may send messages to the appropriate device controller 2510 to activate the requested peripheral equipment or bed system element. For example, based on snoring detection, the bed controller 2504 may send a message to the bed base to adjust the angle of the head or feet, a message to the speaker to start emitting white noise, a message to the pump to adjust the firmness of the bed system, a message to the foot heating controller to activate the foot heater, and a message to the white noise controller to adjust the white noise.

[0291] The device controller 2510 can control peripheral devices (2530). For example, the white noise controller may start a script for white noise in the room around the bed and begin emitting white noise.

[0292] Generally, process 2500 can be organized into training time and operational time. Training time may include operations commonly used to create a snoring classifier, while operational time may include operations commonly used to determine snoring conditions using the classifier. Depending on the configuration of the bed system, one or both of the operations during this time may be operational or suspended. For example, when a user purchases a new bed, the bed may not have access to pressure readings generated by the user on the bed, or to acoustic readings generated by a user snoring. When a significant cluster of readings is collected as the user begins using the bed for the first few nights (e.g., a certain number of readings, a certain number of nights, a certain number of expected in-and-out events based on different tests or heuristics), the bed system may collect those pressure and acoustic readings and provide them to the cloud reporting service 2506.

[0293] The bed system may operate during training time to update or expand classifiers. The bed controller 2504 may continue operating during training time after receiving classifiers. For example, the bed controller 2504 may periodically send pressure and acoustic readings to the cloud reporting service 2506, at user instruction or when computing resources are available. The classifier factory 2508 may generate and send new or updated classifiers, or send a message indicating that one or more classifiers on the bed controller 2504 should be decommissioned.

[0294] The bed controller 2504 can receive rules and settings that define how the home automation connected to the bed system should operate. Using a classifier, the bed system can perform operations during the operation time to cause the home automation to execute according to the rules and settings.

[0295] The bed system can operate simultaneously during the training time and the operation time using the same pressure readings from the pressure sensor 2502 and the same acoustic readings from the acoustic sensor 2502. For example, the bed system can use the streams of pressure readings and acoustic readings to determine the snoring state and control the environment based on the currently used snore classifier. Further, during the operation of the training time, the bed system can improve the classifier using the same pressure / acoustic readings from the stream of pressure / acoustic readings. Thus, a single stream of pressure and acoustic readings can be used both to improve the functions of the bed system and to drive automation events.

[0296] In some cases, a general set of classifiers can be used instead of or in combination with the personalized classifier. For example, when the bed is newly purchased or reset to the factory - shipped settings, the bed system can operate with a general or default snore classifier created based on group - level pressure and acoustic readings rather than at the individual level. That is, before the bed system has the opportunity to learn about specific pressure readings related to a particular user, general classifiers can be created for use in the bed system. These general classifiers can be generated using machine - learning techniques such as those described herein for group - level training data. These general classifiers can alternatively be generated using non - machine - learning techniques. For example, the classifier can include thresholds (e.g., pressure, pressure change over time), and acoustic readings above the threshold can be used to determine one snoring state, and acoustic readings below the threshold can be used to determine another snoring state.

[0297] While a specific number, order, and arrangement of elements are described here, other alternatives are possible. For example, although the generation of classifier 2520 is described as being carried out by classifier factory 2508, the classifier may instead or additionally be generated by bed controller 2506 and may not report pressure and acoustic data to the cloud service.

[0298] In some implementations, the bed system may accommodate two users. In such cases, process 2500 may be adapted in one or more ways to accommodate two users. For example, for each user, the bed system may use two sets of classifiers (some classifiers may be present in both sets simultaneously, and others may not). For example, one set may be used when one side of the bed is occupied, and the other set may be used when the other side of the bed is occupied. This may be useful, for example, when the presence or absence of a second user affects the pressure reading on the first user's side of the bed.

[0299] In some cases, users may want to control their home automation environment based on whether both users are snoring in bed. For example, a rule might specify that white noise should only be activated when only one user is snoring in bed, and not when both users are snoring in bed.

[0300] This example is shown for a single bed controller 2504 that provides pressure / acoustics (2516) and subsequently receives a classifier (2524). However, it will be understood that this system is applicable to many further beds and bed controllers. For example, pressure / acoustics may be received from many bed controllers (e.g., hundreds of thousands), and training data may be synthesized from these many beds to provide data on bed usage by many users. The classifier can then be distributed to some, none, or all of the beds that provided the training data. For example, some beds may receive updated software with a new classifier. Or, as another example, the new classifier may be included only in newly manufactured beds. Or, as yet another example, each bed may receive a classifier tailored specifically to the user of that particular bed.

[0301] Figure 26 is a flowchart of an exemplary process 2600 for training a classifier on pressure and / or acoustic signals. This process 2600 may be used, for example, as part of a process (2520) for generating a classifier.

[0302] A set of features is determined (2602). For example, raw pressure data may be separated into rolling windows of pressure data, and raw acoustic data may be separated into rolling windows of acoustic data. In one example, each window represents one second of data and has 100 readings. In one example, pressure data uses a first-length window and acoustic data uses a second-length window. A vector may be created, and the first 100 fields of the vector may be used to sequentially store each of the 100 readings. Additional fields of the vector are used to store data calculated from the 100 readings. For example, one field may be used to store the amplitude of the spectral peak corresponding to the pressure / acoustic data stream. This value may be used as an approximate proxy for the presence of snoring, with higher amplitudes indicating the presence of snoring. Another field may be used to store the maximum difference between values ​​in the pressure / acoustic data stream. This may indicate the stability level of the readings. Furthermore, several features can be created without a clear or logical explanation beyond their mathematical determination. For example, the count of odd or even readings can be stored in one field of a vector. These fields can be defined by human design or generated by a program.

[0303] The training data is mapped to kernel space (2604). For example, a vector can be mapped to a higher-dimensional space. This higher-dimensional space may have the same number of dimensions as the fields of the vector, or a subset of the N fields of the vector may be used so that the vector is mapped to an N-dimensional space. A kernel function can be found that can partition the space into multiple partitions, each having one cluster of vectors. For example, in a 2D space, a vector may be mapped to one cluster around coordinates [1,1] and another cluster around coordinates [100,100]. The decision boundary y = l00 - x can partition the space such that one cluster is entirely above the line of the function and one cluster is entirely below the line of the function.

[0304] Finding a kernel function can be an automated process or it can involve human interaction. For example, a Monte Carlo process may be used to search for a kernel function in an automated process. In a human-involved process, the computer may present a set of two-dimensional views of vectors to the human, who may create two-dimensional functions to partition the two-dimensional space, and the computer system may construct higher-dimensional functions from these two-dimensional functions.

[0305] The classifier is trained on the mapped feature set (2606). The training data can now be trained using the clustered feature set to identify which cluster represents a particular state. In some cases, this can be supervised training, in which a human can identify the clusters and label each cluster. For example, each time window may be tagged by a different process to identify the snoring state at the time the pressure and acoustic readings for that time window were generated. In some cases, explicit tests may be performed to generate data. Records of known snoring may be sampled, and a person lying in bed snoring may be measured. The logs from this test session may be annotated for various snoring states, and the pressure and acoustic data may be appropriately labeled.

[0306] In some cases, other state identification processes may be used. For example, threshold analysis may be used to generate reliable state annotations, but such analysis may require significantly longer periods of pressure and acoustic data (from several minutes to several hours). In such cases, threshold analysis may be performed on historical pressure and acoustic data to label the snoring state in the pressure and acoustic data. This historical analysis can be performed retrospectively and may not be useful, or not very useful, for real-time snoring state determination for purposes such as home automation, but it may be useful for that purpose. That is, an analysis that requires 30 minutes of data to make a determination may be used here even if it results in an unacceptable 30-minute delay for adjusting bed firmness or raising the head of an adjustable base.

[0307] In some cases, training can be unsupervised training. For example, training may be performed solely by analyzing pressure or acoustic data, without providing any external intelligence. This may include unsupervised clustering of data. Clustering techniques include, but are not limited to, k-means clustering, mixed modeling, hierarchical clustering, self-organizing mapping, and hidden Markov modeling. This may additionally or alternatively include unsupervised labeling of data. For example, instead of training data with a given number of states in a given set, the supervision could generate several clusters and use that number of clusters to determine the number of possible states. These states may be given unique identifiers that do not have specific meaning (e.g., cluster 1, cluster 2, state A, state B). Then, once the supervision is complete, a human may analyze the state information and determine meaningful labels for the states.

[0308] Figure 27 is a swimlane diagram of an exemplary process 2700 for training and using a machine learning classifier to determine and classify snoring events in a bed. Unlike process 2500, process 2700 includes the step of generating a classifier 2702 using a deep learning style of machine learning. In the illustrated example, a deep neural network (DNN), which is a computer model (rather than an organic brain), is described. However, as can be understood, other types of artificial neural networks and / or other types of deep learning (e.g., convolutional neural networks, recurrent neural networks, long-short-term memory - LSTM, etc.) may also be used in process 2700. Furthermore, it will be understood that other types of machine learning may also be used in processes 2500, 2700 to generate classifiers (2520, 2702).

[0309] Generally, in classifier generation (2702), the classifier factory 2508 receives labeled training data from a cloud reporting service. However, unlike process 2500, explicit features are not created as a standalone process. Instead, classifier training operates directly on the labeled training data, rather than on features created from the labeled training data.

[0310] The classifier factory 2508 generates classifiers from pressure / acoustic readings 2702. For example, the classifier factory 2508 may employ artificial neural network machine learning to generate classifiers. The classifier factory 2508 may train the classifiers by first acquiring a large set of variation patterns of pre-classified readings. For example, one or many beds may report reading data to the cloud reporting service 2506. This reading data may be labeled, recorded, and stored for analysis in creating pressure classifiers for use by bed controllers 2504 and / or other bed controllers.

[0311] Tagged data is provided to one or more DNN trainers. The DNN trainers generate an initial DNN by arranging groups of artificial neurons in layers and connecting the output of one layer to the input of another. Generally speaking, these artificial neurons are computer-operable functions that take several inputs, perform their functions, and produce an output. Often, these functions are defined based on a two-part mathematical function—first several linear combinations are performed, followed by a nonlinear function (also called an activation function). However, as can be understood, any technically appropriate function can be used.

[0312] All neurons in a layer are grouped together, and the output of each neuron in that layer is provided as input to the neurons in the next layer. The number of connections between layers is a function of the number of inputs for each neuron within that layer. For example, in a network where each layer has 10 neurons and each neuron has 3 inputs, the network may have 30 (i.e., 10 × 3) connections between one layer and the next. The number of layers, the number of neurons per layer, and the number of inputs per neuron are parameters that the classifier factory 2508 can adjust during the process of initializing the DNN. For example, a network may have tens of layers, each layer may have hundreds of neurons, and each neuron may have tens of inputs. Greater or fewer complexities (number of layers, neurons, and / or inputs) are possible.

[0313] Each connection from the output of one neuron to the input of the next neuron is assigned a weight value. This weight value is initialized, for example, to a random (or pseudo-random) number, or by selecting from a list of possible weights. When the output of one neuron is passed to the input of the next neuron, its value is adjusted by this weight value. For example, the weight value can be a number in the range of 0 to 1, and the value passed is multiplied by this weight value.

[0314] Once such an initial DNN is generated, it can receive training data and operate on that training data. That is, the training data, stored on disk as an ordered sequence of binary data, can be provided as input to the head of the DNN (i.e., the original input neurons of the first layer of the DNN). As can be understood, once the input is provided to the first layer of the DNN, the DNN executes the neurons of that layer and produces an output in the form of a second ordered sequence of binary data. Here, the second ordered sequence of binary data can be interpreted as a classification with a confidence score—that is, the output tells the reader which state (e.g., no snoring, left-side user snoring, right-side user snoring) the DNN classified the data to, along with a confidence score between 0 and 1.

[0315] Using the generated initial DNN, the classifier factory 2508 may refine the DNN to improve the classification results produced by it. To do so, the classifier factory 2508 may calculate a loss function and iteratively modify the DNN until the DNN passes (satisfies) tests such as falling below a threshold or failing to improve through iterative refinement.

[0316] The loss function can be chosen to define how well the DNN classified the tagged training data samples. In the example of confidence levels, where the values ​​are between 0 and 1, a loss function could be used that assigns a loss value of 1 to inaccurate classifications and a loss value (l-confidence level) to correct classifications. In this way, inaccurate classifications provide the largest loss value, while correct classifications provide a small loss when the confidence level is high.

[0317] The classifier factory 2508 begins refining the DNN to reduce its loss value. For example, the classifier factory 2508 may iteratively perform the steps of i) tuning the DNN, ii) providing training data to the DNN, and iii) calculating the DNN's loss value.

[0318] To tune a DNN, the classifier factory can select one or more optimization algorithms. Generally, many of these algorithms work by adjusting the weights of the connections between neuron outputs and neuron inputs. In doing so, they adjust the actual weighted inputs used by the neurons in the DNN. This leads to different results for the DNN.

[0319] One of these algorithms is called the gradient descent algorithm. Gradient descent is a first-order iterative optimization algorithm for finding the minimum value of a loss function. In each iteration of gradient descent, the current weights of the connections between neurons are considered and modified to reduce the DNN's loss value by at least a small amount. To make these modifications, the classifier factory 2508 can determine the gradient of the DNN's loss function for all of the DNN's weights. Using this gradient, new weights are computed that can reduce the loss function by the learning rate. The gradient descent algorithm can also incorporate elements to avoid being trapped at local minima. Exemplary elements include stochastic gradient descent, batch gradient descent, and minibatch gradient descent.

[0320] Once the DNN is tuned, the classifier factory 2508 can generate one or more classifiers from the DNN. For example, the classifier factory 2508 can identify and remove neurons whose all input weights are zero because they do not contribute to the classification performed by the DNN. The claims at the time of filing are as follows: <Claim 1> It is a bed system, A mattress that supports the left user on the left side and the right user on the right side, At least one acoustic sensor configured to sense acoustic energy in the environment for the left user and the right user, At least one pressure sensor configured to sense pressure, which is applied to the left side by a left-side user and to the right side by a right-side user, Controller and Equipped with, At least some of the acoustic energy in the environments of the left user and the right user is created by the left user. At least some of the acoustic energy in the environments of the left user and the right user is created by the right user. The aforementioned controller, Receiving at least one acoustic stream from the at least one acoustic sensor, Receiving at least one pressure stream from the at least one pressure sensor, From the at least one acoustic stream and the at least one pressure stream, left-side snoring / breathing parameters for the left-side user and right-side snoring / breathing parameters for the right-side user are generated. Determine that the home automation rule includes a condition that includes at least one of the left-side snoring / breathing parameter and the right-side snoring / breathing parameter. In response to a determination that the home automation rule includes a condition that includes at least one of the left-side snoring / breathing parameter and the right-side snoring / breathing parameter, a command to drive the controllable device is transmitted to the controllable device. It is configured in such a way, The at least one acoustic stream represents the acoustic energy sensed by the acoustic sensor, The at least one pressure stream represents the pressure sensed by the pressure sensor. A bed system characterized by the following features. <Claim 2> The system further includes the controllable device, The controllable device is Upon receiving the command to drive the controllable device, In response to receiving the command to drive the controllable device, it drives the device to change the user's environment. It is configured in such a way The bed system according to feature 1. <Claim 3> The aforementioned at least one acoustic sensor has two acoustic sensors. The bed system according to claim 1 or 2, characterized in that it is the bed system according to claim 1 or 2. <Claim 4> The aforementioned at least one pressure sensor has two pressure sensors. A bed system according to any one of claims 1 to 3. <Claim 5> The at least one pressure sensor has a configuration that separately senses the pressure applied to the left side by a left-side user and separately senses the pressure applied to the right side by a right-side user. A bed system according to any one of claims 1 to 4. <Claim 6> The at least one acoustic sensor is Signal regulator and, Digitizer and, It has, The aforementioned acoustic stream is a digital data stream. A bed system according to any one of claims 1 to 5. <Claim 7> The at least one pressure sensor is Signal regulator and, Digitizer and, It has, The aforementioned pressure stream is a digital data stream. A bed system according to any one of claims 1 to 6. <Claim 8> The controller further, In order to create a left-side combined stream, one of the one or more acoustic streams is combined with one of the one or more pressure streams, In order to create a right-side combined stream, one of the one or more acoustic streams is combined with one of the one or more pressure streams, From the at least one acoustic stream and the at least one pressure stream, the left-side coupled stream and the right-side coupled stream are used to generate left-side snoring / breathing parameters for a left-side user and right-side snoring / breathing parameters for a right-side user. It is configured in such a way A bed system according to any one of claims 1 to 7. <Claim 9> The controller further, Using one sensing modality, left-side snoring / breathing parameters for a left-side user and right-side snoring / breathing parameters for a right-side user are generated from the at least one acoustic stream and the at least one pressure stream. The left-side snoring / respiratory parameters and the right-side snoring / respiratory parameters are confirmed using the second sensing modality. It is configured in such a way A bed system according to any one of features 1 to 8. <Claim 10> The controller further, Filtering at least one of the at least one acoustic stream and the at least one pressure stream. It is configured in such a way A bed system according to any one of claims 1 to 9. <Claim 11> The controller further, Determine that at least one user is present on the mattress, Determine that at least one user is asleep. It is configured in such a way A bed system according to any one of claims 1 to 10. <Claim 12> The generation of the left-side snoring / respiratory parameters and the right-side snoring / respiratory parameters are performed in response to the determination that at least one user is asleep. A bed system according to any one of claims 1 to 11. <Claim 13> The controller further, Using a machine learning classifier, the left-side snoring / breathing parameters for a left-side user and the right-side snoring / breathing parameters for a right-side user are generated from the at least one acoustic stream and the at least one pressure stream. It is configured in such a way A bed system according to any one of claims 1 to 12. <Claim 14> The process involves monitoring the snoring and breathing sounds of two people sharing a bed, The process involves separating the sound source signal of the first sleeper and the sound source signal of the second sleeper from two acoustic recordings captured by two acoustic sensors, via a sound source separation algorithm. A step of associating each of the two separated sound sources with the first sleeper and the second sleeper, using pressure data from the left and right sleepers, Equipped with, The aforementioned process of associating is, A process of matching the signal features of the first separated acoustic signal and the signal features of the second separated acoustic signal with the signal features of the left-side pressure data and the signal features of the right-side pressure data. Includes A method characterized by the following: <Claim 15> It is a bed system, A mattress for supporting the left-side user on the left side and the right-side user on the right side, and means for supporting the mattress, At least one acoustic sensor configured to sense acoustic energy in the environment for the left user and the right user, At least one pressure sensor configured to sense pressure, which is applied to the left side by a left-side user and to the right side by a right-side user, Controller and Equipped with, The aforementioned controller, Receiving at least one acoustic stream from the at least one acoustic sensor, Receiving at least one pressure stream from the at least one pressure sensor, From the at least one acoustic stream and the at least one pressure stream, left-side snoring / breathing parameters for the left-side user and right-side snoring / breathing parameters for the right-side user are generated. Determine that the home automation rule includes a condition that includes at least one of the left-side snoring / breathing parameter and the right-side snoring / breathing parameter. In response to a determination that the home automation rule includes a condition that includes at least one of the left-side snoring / breathing parameter and the right-side snoring / breathing parameter, a command to drive the controllable device is transmitted to the controllable device. It is configured in such a way, The at least one acoustic stream represents the acoustic energy sensed by the acoustic sensor, The at least one pressure stream represents the pressure sensed by the pressure sensor. A bed system characterized by the following features. <Claim 16> The system further includes the controllable device, The controllable device is Upon receiving the command to drive the controllable device, In response to receiving the command to drive the controllable device, it drives the device to change the user's environment. It is configured in such a way The bed system according to claim 15. <Claim 17> The controller further, In order to create a left-side combined stream, one of the one or more acoustic streams is combined with one of the one or more pressure streams, In order to create a right-side combined stream, one of the one or more acoustic streams is combined with one of the one or more pressure streams, From the at least one acoustic stream and the at least one pressure stream, the left-side coupled stream and the right-side coupled stream are used to generate left-side snoring / breathing parameters for a left-side user and right-side snoring / breathing parameters for a right-side user. It is configured in such a way The bed system according to claim 15 or 16, characterized in that it is the bed system according to claim 15 or 16. <Claim 18> The controller further, Using one sensing modality, left-side snoring / breathing parameters for a left-side user and right-side snoring / breathing parameters for a right-side user are generated from the at least one acoustic stream and the at least one pressure stream. The left-side snoring / respiratory parameters and the right-side snoring / respiratory parameters are confirmed using the second sensing modality. It is configured in such a way A bed system according to any one of claims 15 to 17. <Claim 19> The controller further, Determine that at least one user is present on the mattress, Determine that at least one user is asleep. It is configured in such a way A bed system according to any one of claims 15 to 18. <Claim 20> The generation of the left-side snoring / respiratory parameters and the right-side snoring / respiratory parameters are performed in response to the determination that at least one user is asleep. The bed system according to any one of claims 15 to 19. <Claim 21> A system equipped with sensors and configured to sense parameters of two sleepers on a bed, and to analyze the sensed parameters in order to determine which of the two sleepers is associated with one or more of the sensed parameters.

Claims

1. It is a system, A mattress that supports the left user on the left side and the right user on the right side, At least two acoustic sensors configured to sense acoustic energy in the environment of the left user and the right user, At least one pressure sensor configured to sense pressure, which is applied to the left side by a left-side user and to the right side by a right-side user, A controller having memory and at least one processor, Equipped with, At least some of the acoustic energy in the environments of the left user and the right user is created by the left user. At least some of the acoustic energy in the environments of the left user and the right user is created by the right user. The controller is configured to maintain left / right / no-snoring classifiers in the memory. The aforementioned left / right / no-snoring classifiers include machine learning models generated from a machine learning process on training data. The aforementioned left / right / no-snoring classifier is, It receives data from at least two acoustic streams and at least one pressure stream as input, This function returns a Left / Right / No Snoring object as output, providing a classification from Left-side snoring, Right-side snoring, Both-sided snoring, or No-sided snoring. It is configured in such a way, The controller further, The system receives the at least two acoustic streams from the at least two acoustic sensors. The system receives the at least one pressure stream from the at least one pressure sensor. The data from the at least two acoustic streams and the at least one pressure stream is provided as input to the left / right / no-snoring classifier. It is configured in such a way A system characterized by the following features.

2. The controller is further configured to transmit commands to controllable peripheral devices to drive those peripheral devices in response to the left / right / no-snoring objects. The system according to feature 1.

3. The controller further, From the aforementioned left / right / no snoring objects, the classification of "snoring on the left side" is identified, and this classification of "snoring on the left side" corresponds to the snoring status of the left-side user. From the aforementioned left / right / no snoring objects, the classification of "right-side snoring present" is identified, and this classification of "right-side snoring present" corresponds to the snoring status of the right-side user. It is configured in such a way The system according to claim 1 or 2, characterized in that it is the same as described in claim 1 or 2.

4. The aforementioned left / right / no-snoring object is a data object having a first data field for recording the snoring parameters of the left-side user and a second data field for recording the snoring parameters of the right-side user. The system according to claim 1 or 2, characterized in that it is the same as described in claim 1 or 2.

5. It is a system, A mattress that supports the left user on the left side and the right user on the right side, At least two acoustic sensors configured to sense acoustic energy in the environment of the left user and the right user, At least one pressure sensor configured to sense pressure, which is applied to the left side by a left-side user and to the right side by a right-side user, A second controller having a second memory and at least one second processor, Equipped with, At least some of the acoustic energy in the environments of the left user and the right user is created by the left user. At least some of the acoustic energy in the environments of the left user and the right user is created by the right user. The second controller is configured to maintain the snoring presence classifier in the second memory, The aforementioned snoring presence classifier includes a machine learning model generated from a machine learning process on training data, The aforementioned snoring presence classifier is, The system receives second data from at least two acoustic streams and at least one pressure stream as input. Returns an object representing the existence of snoring as output. It is configured in such a way, The second controller is configured to also maintain the left / right classifiers within the second memory. The aforementioned left / right classifier includes a machine learning model generated from a machine learning process on training data. The aforementioned left / right classifier is, The system receives the second data from the at least two acoustic streams and the at least one pressure stream as input. This function returns left / right objects as output, providing classifications from left-side snoring and right-side snoring, or from left-side snoring, right-side snoring, and bilateral snoring. It is configured in such a way, The second controller further, The system receives the at least two acoustic streams from the at least two acoustic sensors. The system receives the at least one pressure stream from the at least one pressure sensor. The second data from the at least two acoustic streams and the at least one pressure stream is provided as input to the snoring presence classifier. From the object in which snoring is present, it is determined that snoring is present in the second data from the at least two acoustic streams and the at least one pressure stream. In response to the determination that snoring is present in the second data, the second data from the at least two acoustic streams and the at least one pressure stream is provided as input to the left / right classifier. It is configured in such a way A system characterized by the following features.

6. A controller comprising memory and at least one processor, It is configured to maintain left / right / no-snoring classifiers in the memory, The aforementioned left / right / no-snoring classifiers include machine learning models generated from a machine learning process on training data. The aforementioned left / right / no-snoring classifier is, It receives data from at least two acoustic streams and at least one pressure stream as input, This function returns a Left / Right / No Snoring object as output, providing a classification from Left-side snoring, Right-side snoring, Both-sided snoring, or No-sided snoring. It is configured in such a way, The controller further, The system receives the at least two acoustic streams from at least two acoustic sensors, The system receives the at least one pressure stream from at least one pressure sensor, The data from the at least two acoustic streams and the at least one pressure stream is provided as input to the left / right / no-snoring classifier. It is configured in such a way A controller characterized by the following features.

7. The controller is further configured to transmit commands to controllable peripheral devices to drive those controllable peripheral devices in response to the reception of the left / right / no-snoring objects. The controller according to claim 6.

8. The controller further, From the aforementioned left / right / no snoring objects, the classification of "snoring on the left side" is identified, and this classification of "snoring on the left side" corresponds to the presence of snoring in the left-side user. From the aforementioned left / right / no snoring objects, the classification of "right-side snoring present" is identified, and this classification of "right-side snoring present" corresponds to the snoring status of the right-side user. It is configured in such a way The controller according to claim 6 or 7.

9. The aforementioned left / right / no-snoring object is a data object having a first data field for recording the snoring parameters of the left-side user and a second data field for recording the snoring parameters of the right-side user. The controller according to claim 6 or 7.

10. A method for operating a controller having memory and at least one processor, The process of maintaining left / right / no-snoring classifiers in the memory. Equipped with, The aforementioned left / right / no-snoring classifiers include machine learning models generated from a machine learning process on training data. The aforementioned left / right / no-snoring classifier is, It receives data from at least two acoustic streams and at least one pressure stream as input, This function returns a Left / Right / No Snoring object as output, providing a classification from Left-side snoring, Right-side snoring, Both-sided snoring, or No-sided snoring. It is configured in such a way, This method further, A step of receiving at least two acoustic streams from at least two acoustic sensors, A step of receiving at least one pressure stream from at least one pressure sensor, The process involves providing the data from the at least two acoustic streams and the at least one pressure stream as input to the left / right / no-snoring classifier, A method characterized by comprising:

11. The process of transmitting a command to a controllable peripheral device to drive the controllable peripheral device in response to the left / right / no-snoring object. The method according to 10, further comprising the following:

12. A step of identifying the classification of left-side snoring from the aforementioned left / right / no-snoring objects, A step of identifying the classification of "right-side snoring present" from the aforementioned left / right / no-snoring objects, Furthermore, The aforementioned classification of "snoring on the left side" corresponds to the presence of snoring in the left-side user. The aforementioned classification of "right-side snoring" corresponds to the presence of snoring in the right-side user. The method according to 10 or 11, characterized by the features described herein.

13. The aforementioned left / right / no-snoring object is a data object having a first data field for recording the snoring parameters of the left-side user and a second data field for recording the snoring parameters of the right-side user. The method according to 10 or 11, characterized by the features described herein.

Images