An intelligent anti-falling mattress device with active buffering function

Abstract

The intelligent home of the present application relates to the technical field of intelligent home and medical care, in particular to an intelligent anti-falling mattress device with active buffering function, which solves the problems of passive, low precision and the like of the existing bed falling protection. The core innovation is the deep integration of lightweight end-side intelligent algorithm and mechanical execution structure, data is collected through a non-uniform layout flexible pressure sensor array, and a CNN-LSTM-AM hybrid neural network model is relied on to realize real-time inference and decision of the bed falling risk end-side, trigger the electromagnetic buckle to release the spring to pop out the telescopic arm to form a protective barrier, synchronously upload the alarm information, realize the upgrade from passive monitoring to active prediction-immediate protection-synchronous alarm, the end-side algorithm response reaches the level of hundreds of milliseconds, and it is suitable for the elderly and people with difficulty in movement.

Description

Technical Field

[0001] This invention relates to the field of smart home and medical care technology, and in particular to a smart anti-fall mattress device with active cushioning function. Background Technology

[0002] As the population ages, the problem of elderly people and those with mobility issues slipping out of bed and falling while sleeping is becoming increasingly prominent, placing a heavy burden of care on families and society. Currently, existing solutions for preventing falls from beds mainly fall into four categories: First, physical guardrails, which prevent people from slipping by installing guardrails along the edge of the bed. However, this method limits the convenience of getting in and out of bed, and the guardrails themselves pose a risk of secondary injury. Second, wearable monitoring devices, such as smart bracelets, use accelerometers to detect the state of a person falling. However, these devices require the user to wear them actively, are uncomfortable, are easily forgotten, cannot provide 24-hour continuous monitoring, and can only issue an alarm after a fall has occurred, thus providing passive protection. Third, video surveillance systems, which use cameras installed indoors to monitor the bed in real time. However, this method involves user privacy issues, severely limiting its applicability. Fourth, existing smart mattresses, which have built-in pressure sensors to monitor the status of people on the bed. However, the sensor layout of these mattresses is sparse, and they can only simply determine whether the bed is occupied. They cannot accurately detect changes in body posture and pressure distribution trends, so they cannot send timely warning information to family members, nor can they provide early protection against the risk of falling from the bed. Moreover, the overall cost of these devices is high, making them difficult to popularize in ordinary households.

[0003] In view of the shortcomings of the existing technologies, there is an urgent need for a bedside safety device that can identify and actively protect against bed falls in advance, while taking into account monitoring accuracy, ease of use, privacy and universality, so as to solve the safety hazards of bed falls for the elderly and people with limited mobility. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent anti-fall mattress device with active cushioning function.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A smart anti-fall mattress device with active cushioning function includes a cross-shaped active anti-fall structure embedded inside the mattress. The mechanical actuation part of the cross-shaped active anti-fall structure consists of a cross-shaped central body, four telescopic arms, compression springs, and electromagnetic buckles. The four telescopic arms are installed around the cross-shaped central body using a drawer-type slide rail structure. Each telescopic arm is equipped with a compression spring on both sides. In the standby state, the compression springs are locked and compressed by the electromagnetic buckles. When triggered, the electromagnetic buckles pop open, the springs release, and push the telescopic arms out to form a flexible protective barrier.

[0006] The device also integrates a sensing layer, a low-level control layer, a communication layer, a power drive layer, and a human-machine interaction layer. These layers are electrically connected and achieve intelligent collaborative operation throughout the entire process through edge-side algorithms. 1. Sensing layer: A flexible pressure sensor with a hollow head covering that covers the center of the cross. It adopts a non-uniform pressure sensor array design and only places sensing nodes in key areas of human contact to collect continuous temporal and spatial data of human pressure distribution. This provides high-dimensional, non-redundant raw sensing data for edge-side algorithms, balancing monitoring accuracy and hardware cost. 2. Bottom control layer: An Arduino controller embedded in the center of the cross, serving as the core control and algorithm inference unit on the end side of the device. It receives preprocessed pressure data from the perception layer in real time and runs a lightweight optimized CNN-LSTM-AM hybrid neural network model algorithm. Through a three-level inference logic of spatial feature extraction, temporal trend modeling, and high-risk feature enhancement, it achieves accurate and real-time identification of human sleeping posture, slippage trend, and risk of falling out of bed. 3. Communication Layer: This is an IoT communication module that connects to the underlying control layer via a UART serial port. It enables wireless data transmission between the device, the cloud server, and the family member's mobile app. It can simultaneously upload algorithm inference results, risk characteristic data, and other information to complete the real-time reporting of multi-dimensional alarm information and achieve end-to-cloud data linkage. 4. Power drive layer: This layer consists of MOSFETs or relay modules to solve the problem of low output current of the I / O ports of the underlying control layer. It enables precise control of the on / off state of the electromagnetic latch, ensuring that the algorithm decision results are converted into mechanical actions without delay and with reliability. 5. Human-Machine Interaction Layer: This is an independent physical emergency button located on the outer casing of the device. The signal line connects to the interrupt input pin of the underlying control layer, which can forcibly interrupt the normal reasoning process of the algorithm, activate the protection and alarm functions, realize the dual protection logic of automatic algorithm decision-making and manual forced triggering, and meet the active operation needs in special situations.

[0007] The core working principle of this device is an intelligent protection closed loop driven by the entire algorithm process: The bottom control unit collects and preprocesses the voltage changes of the flexible pressure sensor in real time to construct a feature dataset of human body pressure distribution; the dataset is input into the CNN-LSTM-AM hybrid neural network model for end-side inference to accurately determine the sleeping posture and slippage trend of the human body and identify the risk of falling out of bed; when the risk of falling out of bed is determined or the emergency button is detected to be pressed, the bottom control unit outputs a high level to turn on the MOSFET, causing the electromagnetic latch to pop open and release the compression spring, pushing the telescopic arm to pop out and form a bedside protective barrier; at the same time, the bottom control unit uploads the risk information and feature data inferred by the algorithm to the cloud through the Internet of Things communication module, and the cloud pushes it to the family's mobile phone applet to realize the intelligent closed loop of risk prediction-active protection-synchronous alarm.

[0008] The beneficial effects of this invention are as follows: This application combines high-density flexible sensing with the CNN-LSTM-AM hybrid neural network model algorithm of the underlying control unit to upgrade the traditional passive monitoring of bed falls to active identification and real-time protection. It can complete the protection trigger before the bed fall accident occurs, greatly reducing the probability of fall injury. The device adopts the edge local computing mode, and the underlying control unit achieves local response at the level of hundreds of milliseconds. The hardware works together to ensure the timeliness and reliability of protection triggering. The flexible pressure sensor array with a non-uniform layout and a hollow headgear is used. Sensors are placed only in key areas, which can accurately sense the distribution of human body pressure, changes in posture and slippage trend, while avoiding redundant sensor layout. The device is embedded inside the mattress, with no exposed protective structure, so it does not affect the user's normal getting in and out of bed and sleep experience. Moreover, the user does not need to actively wear any device, achieving imperceptible protection. Instead of video surveillance, data is collected via pressure sensors, preventing privacy leaks at the source. Physical protection uses a spring-driven telescopic arm, eliminating the risk of secondary injury from rigid guardrails.

[0009] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of the structure of an intelligent anti-fall mattress device with active cushioning function according to the present invention; Figure 2 This is a schematic diagram of the control unit shown in the present invention.

[0011] Reference numerals: 1-Cross center body, 2-Telescopic arm, 3-Compression spring, 4-Electromagnetic buckle, 5-Control unit, 6-Emergency button, 7-Drawer slide rail, 8-Arduino controller, 9-Flexible pressure sensor. Detailed Implementation

[0012] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0014] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0015] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0016] This embodiment provides an intelligent anti-fall mattress device with active cushioning function. The core is a fusion architecture of a cross-shaped active anti-fall structure embedded inside the mattress and a lightweight intelligent algorithm on the end side. Unlike the traditional inventions that only optimize the mechanical structure, this device uses the algorithm as the core to drive the entire process of protection. Its mechanical execution part includes a cross-shaped central body 1, four drawer-type telescopic arms 2, compression springs 3, and electromagnetic buckles 4. The four telescopic arms 2 are installed in the front, back, left, and right directions of the cross-shaped central body 1, respectively. Each telescopic arm 2 has two compression springs 3 on each side. In the standby state, the electromagnetic buckles 4 lock the compression springs 3 in a compressed state. When triggered, the electromagnetic buckles 4 are de-energized and pop open, the springs release elastic potential energy, and push the telescopic arms 2 outward along the drawer-type slide rails 7 to form a flexible protective barrier around the edge of the bed, effectively preventing the human body from sliding off the bed.

[0017] The sensing layer of the device uses a hollow headgear-type flexible pressure sensor 9. This sensor has a non-uniform array layout and only sets sensing nodes in key areas of the human body, such as the head, back, and buttocks, that are in contact with the mattress. This matches the spatial feature extraction logic of the underlying algorithm. The sensor's signal line is connected to the analog input pin of the control unit 5 to collect pressure distribution time-domain data at a sampling frequency of 20Hz.

[0018] The control unit 5 uses an Arduino controller 8 as its bottom control layer, which is embedded in the internal slot of the cross-shaped central body 1, serving as the core carrier for edge algorithm deployment and inference. The control unit 5 receives voltage data transmitted from the sensing layer in real time, first performs raw data preprocessing through sliding filtering and normalization, and then inputs the preprocessed data into a pre-programmed lightweight CNN-LSTM-AM hybrid neural network model. The CNN layer of this model uses 3×3 small convolutional kernels to extract the spatial topological features of pressure distribution, the LSTM layer sets 64 hidden units to perform temporal modeling of continuously sampled data, and captures the dynamic trend of the human body's center of gravity shifting towards the edge of the bed. The AM layer assigns a weight coefficient of 0.8 to the pressure change features in the edge area of ​​the bed to enhance high-risk features. Through three-level inference logic, it accurately judges whether the user has a tendency to slip and the risk of falling off the bed. The total latency of edge algorithm inference + local response of the control unit 5 is controlled within 80ms, achieving a fast decision-making process at the hundred-millisecond level.

[0019] The communication layer uses the ESP8266 Wi-Fi module, which is connected to the TX / RX pins of the control unit 5 via the UART serial port. The module establishes a connection with the home wireless network to achieve wireless communication with the cloud server. When the alarm is triggered, it can simultaneously upload multi-dimensional information such as the risk level, trigger time, and pressure distribution characteristic segments in the algorithm inference. It also supports point-to-point data transmission with family members' mobile phone mini-programs.

[0020] The power drive layer uses an N-channel MOSFET module. The digital output pin D9 of the control unit 5 is connected to the input of the MOSFET module, and the output of the MOSFET module is connected in series in the 12V power supply circuit of the electromagnetic latch 4. Since the output current of the I / O port of the control unit 5 is only 20mA, it cannot directly drive the electromagnetic latch 4. The MOSFET module amplifies the current to 1A to ensure that the electromagnetic latch 4 can reliably open, realizing the zero-delay conversion of the algorithm decision result into the mechanical execution action.

[0021] The human-computer interaction layer consists of a physical emergency button 6 located on the outer shell of the cross-shaped active fall protection structure. The button is designed to be waterproof and prevent accidental touches. Its signal line is connected to the interrupt input pin INT0 of the control unit 5. When the user feels unwell or is at risk of slipping, he / she can manually press the button to directly interrupt the normal reasoning process of the algorithm, force the activation of the protection function and upload alarm information, thus realizing the dual protection logic of automatic algorithm decision-making and manual forced triggering.

[0022] The specific working steps of the device in this embodiment are an intelligent protection closed loop driven by the entire algorithm process: 1. Sensing and Preprocessing Stage: After the device is powered on, the control unit 5 reads the voltage change of the flexible pressure sensor in real time through the analog input pin at a sampling frequency of 20Hz. It collects 20 sets of pressure distribution data per second, removes environmental interference through sliding filtering, normalizes and unifies the data scale, and constructs a feature dataset of human posture and center of gravity changes, continuously providing high-quality sensing data for algorithm inference. 2. Edge-side intelligent reasoning and decision-making stage: The control unit 5 inputs the preprocessed feature dataset into the CNN-LSTM-AM hybrid neural network model. The CNN layer extracts the spatial features of pressure distribution, the LSTM layer captures the temporal offset trend, and the AM layer strengthens the high-risk features of the bed edge. The model's edge-side reasoning time is ≤50ms. When the model detects that the center of gravity of the person has shifted to the bed edge beyond the preset threshold, it determines that there is a risk of falling out of bed. If the emergency button 6 is detected to be pressed, the interrupt program is directly triggered, and an emergency state is determined. 3. Algorithm-driven protection execution phase: When a risk of falling from the bed or an emergency is detected, the digital output pin D9 of the control unit 5 immediately outputs a high level, turns on the MOS transistor module, connects the power supply circuit of the electromagnetic latch 4, and the latch pops open instantly, releasing the compressed spring. The spring pushes the four telescopic arms 2 to pop outward along the slide rail, with a pop-out distance of 30cm, forming a flexible protective barrier beside the bed. This process takes ≤30ms from algorithm decision to execution completion. 4. Multi-dimensional alarm reporting stage: While the high-level output triggers the electromagnetic latch 4, the control unit 5 sends a command to the ESP8266 Wi-Fi module via the UART serial port. The module uploads alarm information such as the risk level, trigger time, and human body pressure distribution characteristic segments inferred by the algorithm to the cloud server. The cloud server immediately pushes the multi-dimensional alarm information to the family member's mobile app. The family member can view the on-site status in real time through the app, realizing synchronous and delay-free protection triggering and alarm uploading.

[0023] After the protection is triggered, the user can manually push the telescopic arm 2 back to its original position. The electromagnetic buckle 4 will automatically reset and relock the compression spring 3, and the device will return to standby mode to continue to collect pressure data and perform algorithm inference monitoring, so as to achieve repeated use.

[0024] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0025] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A smart anti-falling mattress device with active cushioning function, characterized in that, The device includes a cross-shaped active fall prevention structure, which is embedded inside the mattress. Its mechanical actuation components include a central cross body, four telescopic arms, compression springs, and electromagnetic latches. The four telescopic arms are installed around the central cross body using a drawer-type slide rail structure. Each telescopic arm has a compression spring on both sides, which is locked and compressed by the electromagnetic latches in standby mode. The device also integrates a sensing layer, a bottom control layer, a communication layer, a power drive layer, and a human-machine interaction layer. These layers are electrically connected and cooperate with each other. Through real-time reasoning and decision-making by edge-side intelligent algorithms, it achieves accurate monitoring of bed fall risk, rapid protection triggering, and synchronous uploading of alarm information.

2. The intelligent anti-fall mattress device with active cushioning function according to claim 1, characterized in that, The sensing layer is a hollow headgear-type flexible pressure sensor covering the center of the cross. The flexible pressure sensor is a non-uniformly distributed pressure sensor array, with sensing nodes only placed at key contact points on the human head, back, and buttocks. Its signal lines are electrically connected to the underlying control layer to collect continuous time-domain data of human body pressure distribution and transmit it to the underlying control layer, providing high-dimensional and timely raw sensing data for the edge-side algorithm.

3. The intelligent anti-fall mattress device with active cushioning function according to claim 1, characterized in that, The underlying control layer is an Arduino controller embedded in the central body of the cross. The Arduino controller is the core of the entire device for end-side data processing, algorithm reasoning, and logic control. It is pre-programmed with a lightweight and optimized CNN-LSTM-AM hybrid neural network model algorithm. This algorithm extracts the spatial features of pressure distribution data through convolutional neural networks, captures the temporal trend of human posture changes through long short-term memory networks, and combines attention mechanisms to strengthen the weights of high-risk features such as bed edge deviation, so as to achieve real-time and accurate end-side analysis and judgment of human sleeping posture, slippage trend, and bed fall risk.

4. The intelligent anti-fall mattress device with active cushioning function according to claim 1, characterized in that, The communication layer is an IoT communication module connected to the underlying control layer. The IoT communication module is either an ESP8266 Wi-Fi module or a 4G module. It is connected to the TX / RX pins of the underlying control layer via a UART serial port to realize wireless data transmission between the device and the cloud and family members' mobile phone mini-programs. It can synchronously upload information such as algorithm inference results, risk trigger types, and pressure sensing raw data fragments to achieve end-to-cloud data linkage.

5. The intelligent anti-fall mattress device with active cushioning function according to claim 1, characterized in that, The power drive layer includes a MOSFET. The digital output pin of the bottom control layer is connected to the input terminal of the MOSFET. The output terminal of the MOSFET is connected in series in the power supply circuit of the electromagnetic latch to amplify the output current of the bottom control layer, realize precise control of the on and off of the electromagnetic latch, and ensure that the algorithm decision result is converted into mechanical execution action without delay.

6. The intelligent anti-fall mattress device with active cushioning function according to claim 1, characterized in that, The human-machine interaction layer is an independent physical emergency button set on the outer shell of the cross-shaped active fall protection structure. The signal line of the emergency button is directly connected to the interrupt input pin of the underlying control layer. After being triggered, it can directly interrupt the normal reasoning process of the algorithm and forcibly start the protection and alarm functions, meet the needs of manual active operation in special circumstances, and realize the dual protection logic of automatic algorithm decision-making and manual forced triggering.

7. The intelligent anti-fall mattress device with active cushioning function according to any one of claims 1-6, characterized in that, The operation steps of the device are driven by a full-process edge-side algorithm, specifically including: S1. High-efficiency data acquisition: The bottom control unit reads the voltage changes of the hollow head-type flexible pressure sensor in real time at a preset sampling frequency to obtain continuous temporal and spatial data of human body pressure distribution. After denoising and normalizing the raw data, a feature dataset of human posture and center of gravity changes is constructed to provide preprocessed data for algorithm inference. S2, Edge-side intelligent risk decision-making: The underlying control unit inputs the preprocessed feature dataset into a lightweight CNN-LSTM-AM hybrid neural network model. The CNN layer extracts the spatial topological features of the pressure distribution, the LSTM layer performs temporal modeling on the continuously sampled data to capture the dynamic trend of the body's center of gravity shifting towards the edge of the bed, and the AM layer assigns high weights to the pressure change features in the bed edge area. The model quickly determines whether there is a risk of falling out of bed through edge inference; at the same time, it detects in real time whether the emergency button has been pressed, realizing dual risk assessment through automatic algorithm decision-making and manual triggering. S3, Algorithm-driven protection trigger: If the model inference determines that there is a risk of falling from the bed or detects that the emergency button has been pressed, the designated digital output pin of the underlying control unit immediately outputs a high level, turns on the MOSFET or relay module, and allows current to flow through the coil of the electromagnetic latch. The electromagnetic latch pops open instantly, releasing the compressed spring. The spring pushes the telescopic arm to pop out along the drawer-type slide rail, forming a flexible protective barrier next to the bed, realizing the millisecond-level conversion of the algorithm decision result into the mechanical execution action. S4. Multi-dimensional alarm reporting: Simultaneously with the triggering of the electromagnetic latch, the underlying control unit sends a command to the IoT communication module via a serial port, uploading alarm information such as the risk level, trigger time, and human body pressure distribution characteristics determined by the algorithm to the cloud server. The cloud server then accurately pushes the multi-dimensional alarm information to the family member's mobile app, achieving synchronous and time-free triggering of protection and uploading of alarm information.

8. The intelligent anti-fall mattress device with active cushioning function according to claim 7, characterized in that, The total latency of edge algorithm inference and local response of the underlying control unit is in the hundreds of milliseconds, of which the edge inference latency of the CNN-LSTM-AM hybrid neural network model is ≤50ms, realizing a high-speed response throughout the entire process of rapid identification of bed fall risk, algorithm decision-making and protection triggering.

Images