Electronic aerosol delivery system including motion sensors and AI systems

Abstract

The electronic aerosol delivery system includes a motion sensor configured to provide a data sample regarding the motion of at least a portion of the electronic aerosol delivery system, a filter configured to extract features from the motion sensor data sample, and an artificial intelligence (AI) system configured to receive the extracted features and use the extracted features to identify different user inputs to the electronic aerosol delivery system.

Description

[Technical Field]

[0001] The present disclosure relates to electronic aerosol delivery systems. [Background technology]

[0002] Electronic aerosol delivery systems (devices), including e-cigarettes, electronic vapor delivery devices and systems, electronic aerosol / vapor / nicotine delivery devices and systems, etc., can have a modular format. For example, such devices (systems) can include a cartridge containing an aerosol precursor material, such as a liquid reservoir, and a control unit including a power source, such as a battery. When a user operates the device, such as by pressing a button or inhaling on the device's mouthpiece, the control unit operates the battery to provide power to generate an aerosol from the aerosol precursor material. In many devices, the cartridge includes an atomizer, such as a resistive heater, that generates an aerosol by vaporizing a small amount of liquid (such cartridges are sometimes called cartomizers).

[0003] Thus, electronic aerosol delivery systems typically incorporate two consumables: first, liquid or other aerosol precursor material, and second, battery power. With regard to the former, after the reservoir of liquid or other aerosol precursor material is depleted, the cartridge can be refilled or, alternatively, discarded so that it can be replaced with a new cartridge. With regard to the latter, e-cigarettes typically include some form of wired or wireless (inductive) capability to receive power from an external charging facility, thereby allowing the battery to be recharged.

[0004] Electronic aerosol delivery systems sometimes have more sophisticated features. For example, some systems may provide a user control interface for modifying the level, duration, and / or time profile of heating power delivered by the battery. Such modifications may be useful for personalizing the system for a particular user (or the mood of a particular user). Another example of a user-controlled operation is the entry of a personal identification number (PIN), which may be required to enable use of the device.

[0005] However, while it is desirable for electronic aerosol delivery systems to have user interfaces that support such increasingly complex functionality, it is still desirable to provide electronic aerosol delivery systems that are small, easily portable, robust, consume low power, and are not too expensive. Reconciling these various design goals can be challenging for developers of electronic aerosol delivery systems. Summary of the Invention

[0006] The present disclosure is defined in the appended claims.

[0007] Provided herein is an electronic aerosol delivery system comprising: a motion sensor configured to provide data samples relating to the motion of at least a portion of the electronic aerosol delivery system, a filter configured to extract features from the motion sensor data samples, and an artificial intelligence (AI) system configured to receive the extracted features and use the extracted features to identify different user inputs to the electronic aerosol delivery system.

[0008] Also provided herein is a method of operating an electronic aerosol delivery system, the method including obtaining data samples relating to the movement of at least a portion of the electronic aerosol delivery system from a motion sensor, filtering the data samples to extract features of the data samples, and using an artificial intelligence (AI) system to identify different user inputs to the electronic aerosol delivery system from the features. [Brief explanation of the drawings]

[0009] Various implementations of the invention will now be described in detail, by way of example only, with reference to the following drawings, in which: [Figure 1] 1 is a high-level schematic (exploded view) of an electronic aerosol delivery system (device). [Figure 2] FIG. 2 is a high-level schematic diagram of the control unit of the electronic aerosol delivery system of FIG. 1. [Figure 3] FIG. 2 is a high-level schematic diagram of a cartomizer (cartridge) of the electronic aerosol delivery system of FIG. 1. [Figure 4] FIG. 3 is a high-level schematic diagram of certain electrical components of the control unit of FIG. 2, including an artificial intelligence (AI) system. [Figure 5] FIG. 2 is a high-level schematic diagram illustrating an example of using an AI system of the electronic aerosol delivery system of FIG. 1 to recognize and output recognized gestures. [Figure 6] FIG. 1 illustrates an exemplary data structure in table form for storing data samples corresponding to spatial motion of an electronic aerosol delivery system. [Figure 7] FIG. 10 illustrates an exemplary data structure in table form for storing extracted features corresponding to spatial motion of an electronic aerosol delivery system. [Figure 8] FIG. 1 illustrates an exemplary data structure in the form of a lookup table for storing data used by an AI system to identify user input. [Figure 9]FIG. 2 is a high-level schematic diagram illustrating an example of using the AI ​​system of the electronic aerosol delivery system of FIG. 1 to recognize and output identified user input. [Figure 10] 2 is a schematic flow chart illustrating a process for identifying user input according to the electronic aerosol delivery system of FIG. 1. [Figure 11] 2 is a schematic flow chart illustrating a process for identifying user input according to the electronic aerosol delivery system of FIG. 1. DETAILED DESCRIPTION OF THE INVENTION

[0010] The present disclosure relates to electronic aerosol delivery systems (devices). As used herein, the term "electronic aerosol delivery system" refers to an aerosol delivery system that includes one or more electronic components, such as a controller for controlling the operation of the electronic aerosol delivery system. The electronic aerosol delivery system may or may not include its own power source (such as a battery). The controller may be configured to control any suitable operation of the aerosol delivery device. Operation includes, but is not limited to, delivering at least one substance to a user. Generation of the aerosol from the aerosol-generating material may or may not be achieved by electronic means.

[0011] In some implementations, the electronic aerosol delivery system is a "non-combustible" aerosol delivery system. According to the present disclosure, a "non-combustible" aerosol delivery system is one in which the constituent aerosol-generating materials of the aerosol delivery system (or components of the aerosol delivery system) are not combusted or are not combusted to facilitate the delivery of at least one substance to a user.

[0012] In some implementations, the non-combustible aerosol delivery system is an electronic cigarette (e-cigarette), also known as a vaping device or electronic nicotine delivery system (END), although it should be noted that the presence of nicotine in the aerosol-generating material is not a requirement.

[0013] In some implementations, the non-combustion aerosol delivery system is an aerosol-generating material heating system, also known as a non-combustion heating system, an example of such a system being a tobacco heating system.

[0014] In some implementations, the non-combustible aerosol delivery system is a hybrid system that generates an aerosol using a combination of aerosol-generating materials, where one or more of the aerosol-generating materials can be heated. Each of the aerosol-generating materials can be, for example, in solid, liquid, or gel form and may or may not contain nicotine. In some implementations, the hybrid system includes a liquid or gel aerosol-generating material and a solid aerosol-generating material. The solid aerosol-generating material may include, for example, tobacco or a non-tobacco product.

[0015] Typically, a non-combustible aerosol delivery system may include a non-combustible aerosol delivery device and a consumable item for use with the non-combustible aerosol delivery device.

[0016] In some implementations, consumables that include or are configured with aerosol-forming materials are configured for use with non-flammable aerosol delivery devices. These consumables may be referred to as articles throughout this disclosure.

[0017] In some implementations, the non-combustible aerosol delivery system can include a heat-generating power source, which can include a carbon substrate that can be activated to deliver power in the form of heat to the aerosol-generating material or a heat transfer material in proximity to the heat-generating power source.

[0018] In some implementations, the non-flammable aerosol delivery system may include an area for receiving consumables, an aerosol generator, an aerosol generation area, a housing, a mouthpiece, a filter, and / or an aerosol modifier.

[0019] In some implementations, consumables for use with non-combustible aerosol delivery devices may include an aerosol-generating material, an aerosol-generating material storage area, an aerosol-generating material delivery component, an aerosol generator, an aerosol-generating area, a housing, a wrapper, a filter, a mouthpiece, and / or an aerosol modifier.

[0020] In some implementations, an electronic aerosol delivery system may comprise a combustible aerosol delivery system. According to this disclosure, a "combustible" aerosol delivery system is one in which constituent aerosol-generating materials of the aerosol delivery system (or components of the aerosol delivery system) are combusted or burned during use to facilitate delivery of at least one substance to a user.

[0021] As used herein, an aerosol-generating material is a material that can generate an aerosol when activated, for example, by heating, irradiation, or any other means. The aerosol-generating material can be in the form of a solid, liquid, or gel, which may or may not contain, for example, an active substance and / or a flavoring. In some implementations, the aerosol-generating material can include an "amorphous solid," which may alternatively be referred to as a "monolithic solid" (i.e., non-fibrous). In some implementations, the amorphous solid can be a dried gel. An amorphous solid is a solid material that can retain some fluid, such as a liquid, within it. In some implementations, the aerosol-generating material can include, for example, about 50%, 60%, or 70% by weight of an amorphous solid to about 90%, 95%, or 100% by weight of an amorphous solid.

[0022] Optionally, the aerosol-generating material may include one or more active ingredients, one or more flavorings, one or more aerosol-forming materials, and / or one or more other functional materials. As used herein, an active substance may be a physiologically active material, which is a material intended to achieve or enhance a physiological response. The active substance may be selected from, for example, dietary supplements, nootropics, and psychotropic agents. The active substance may be naturally derived or synthetically obtained. The active substance may include, for example, nicotine, caffeine, taurine, theine, vitamins such as B6, B12, or C, melatonin, cannabinoids, or constituents, derivatives, or combinations thereof. The active substance may include one or more constituents, derivatives, or extracts of tobacco, cannabis, or another botanical substance. In some implementations, the active substance includes nicotine. As used herein, the terms "flavor" and "flavoring" refer to materials that may be used, where local regulations permit, to create a desired taste, aroma, or other somatosensory sensation in products intended for adult consumers. These may include naturally derived flavor materials, botanicals, extracts of botanicals, synthetically derived materials, or combinations thereof. The aerosol-forming material may include one or more constituents capable of forming an aerosol, such as glycerin or glycol. The one or more other functional materials may include one or more of a pH adjuster, colorant, preservative, binder, filler, stabilizer, and / or antioxidant.

[0023] FIG. 1 is a schematic (exploded view) of an example electronic aerosol delivery system. The system is generally cylindrical, extends along a longitudinal axis indicated by dashed line LA, and includes two main components: a control unit (body) 20 (sometimes referred to herein as an (electronic) aerosol delivery device (or, more simply, device), which is generally a reusable component) and a cartomizer (cartridge) 30 (usually representing a consumable component). The aerosol delivery device 20 and the consumable 30 together form the aerosol delivery device 20. The system 10 is generally compact for easy portability (e.g., in a pocket or bag) and handheld use.

[0024] The cartomizer 30 includes an aerosol-generating material storage area (which in this example is an internal chamber containing a reservoir of liquid (liquid being an example of an aerosol-generating material)), an aerosol generator (sometimes called a vaporizer) (which in the following example is a heater), and a mouthpiece 35. However, it should be understood that, consistent with the above, different aerosol-generating materials other than liquids may be used.

[0025] In some implementations, the liquid in the reservoir typically includes nicotine in a suitable solvent and may include additional constituents, for example, to aid in aerosol formation and / or for additional flavoring, as described above. The reservoir may include a foam matrix or any other structure for retaining the liquid until it is delivered to the vaporizer. Alternatively, the liquid may be freely held in the reservoir. The cartomizer 30 may further include a wick or similar feature for transporting a small amount of liquid from the reservoir to a heated location adjacent to the heater (more generally, a wick is an example of an aerosol-generating material delivery component).

[0026] Control unit 20 typically includes at least one rechargeable cell or battery for powering system 10 and at least one circuit (e.g., provided as a printed circuit board (PCB) or flexible circuit) for overall system control. When the heater receives power from the battery as controlled by the circuit board, it vaporizes liquid from the wick, and this vapor is then inhaled by the user through mouthpiece 35. This use of an electronic aerosol delivery system, in which the user inhales electrically generated vapor through a mouthpiece, is typically referred to as vaping.

[0027] The control unit 20 and the cartomizer 30 can be separated from each other by separating them in a direction parallel to the longitudinal axis (LA) of the aerosol delivery device 20, as shown in FIG. 1, but are joined together for use by connections shown schematically as 25A and 25B in FIG. 1. This may be implemented as a bayonet or threaded joint or any other suitable type of coupling. Thus, the control unit 20 can be said to include an area or region for receiving consumables. The connections 25A, 25B provide mechanical and electrical connectivity between the control unit 20 and the cartomizer 30. The control unit 20 may also be provided with a feature (not shown) for connecting the control unit to an external power source. For example, this feature may comprise a (micro / mini / Type-C) USB port.

[0028] The system 10 may be provided with one or more external ports (not shown in FIG. 1 ) for air intakes. These ports may be located in the control unit 20 and may connect via connectors 25A, 25B to an air passage through the control unit, and then through the cartomizer 30 to the mouthpiece 35. When a user inhales through the mouthpiece 35, air is drawn into the control unit, and this airflow (or resulting change in pressure) may be detected by a pressure sensor. In response to this detection, the system may activate a heater to vaporize the liquid received from the reservoir (via the wick). The airflow through the vaporizer combines with the resulting vapor, and this combination of airflow and vapor exits the cartomizer 30 through the mouthpiece 35 for inhalation by the user. When the liquid supply is depleted, the cartomizer 30 may be detached from the main body 20, discarded, and replaced with another cartomizer as needed. In some implementations, the cartomizer is alternatively (or additionally) refillable. Thus, the liquid represents an aerosol-forming material for use with device 20 .

[0029] FIG. 2 is a simplified schematic diagram of the control unit 20 of the electronic aerosol delivery system of FIG. 1, which can be generally considered a cross-section in a plane containing the longitudinal axis LA. As shown in FIG. 2, the control unit 20 includes a battery 210 and a printed circuit board 202 carrying at least one chip, such as an application-specific integrated circuit (ASIC) or microcontroller, for controlling the system 10. The PCB 202 can be located to the side of the battery 210 or at one end. In the configuration shown in FIG. 2, the PCB is located between the battery 210 and the connector 25B. The control unit may also include airflow and / or pressure sensors (not shown) used to detect (among other things) inhalation at the mouthpiece 35. In response to such detection of inhalation, the sensors notify the chips on the PCB 202, which in turn initiate the flow of power from the battery 210 to the cartomizer heater. The control unit 20 may include one or more inlet holes (not shown) to allow air to enter the control unit 20 and flow past the sensors when a user inhales at the mouthpiece 35. This allows the sensor to detect the user's inhalation.

[0030] The distal end of the control unit 20 (i.e., the end opposite the mouthpiece 35 when the system 10 is in use) is designated as the tip end 225, and the opposite end of the control unit 20 (i.e., the proximal end closest to the user during use) is the connector 25B for mating the control unit 20 to the cartomizer 30. As previously mentioned, the connector 25B provides mechanical and electrical connectivity between the control unit 20 and the cartomizer 30. As shown in FIG. 2 , the connector 25B may include a body (control unit) connector 240. The body connector 240 may be metallic (or metal-coated) and serve as a first (outer) terminal for electrical connection (positive or negative) to the cartomizer 30. The connector 25B further includes an electrical contact 250 that provides a second (inner) terminal for electrical connection to the cartomizer 30 that is opposite in polarity to the first terminal. The body connector 240 generally has an annular or tubular shape aligned with the longitudinal axis LA of the control unit 20 (and the entire system 10). The electrical contact 250 may be in the form of a pin that is centrally located in the body connector 240. That is, the contact 250 is aligned and coincident with the longitudinal axis LA. The body connector 240 and the electrical contact 250 are separated by an insulator 260 that is also annular in shape.

[0031] FIG. 3 is a schematic diagram of the cartomizer 30 of the system 10 of FIG. 1, again generally considered as a cross-section in a plane containing the longitudinal axis LA. The cartomizer 30 includes an inner tube 31 that provides and surrounds an air passageway 355 that extends along the central (longitudinal) axis of the cartomizer 30 from the mouthpiece 35 to a connector 25A for connecting the cartomizer to the control unit 20. A reservoir 360 of liquid (typically containing nicotine in a solvent) is provided around the air passageway 355. For example, the reservoir 360 may be formed between the tube defining the air passageway 355 and the outer housing of the cartomizer 30. The reservoir 360 may comprise cotton or foam soaked in the liquid, or the liquid may be freely retained in the reservoir 360 (i.e., without such cotton, foam, or other retention matrix). The liquid acts as an aerosol precursor material, as described in more detail below.

[0032] The cartomizer further includes a mechanical and electrical connector 25A for coupling to the mechanical and electrical connector 25B of the control unit 20. The connector 25A is complementary in shape and structure to the connector 25B and includes an inner electrode 375 and an outer electrode 370 separated by an insulator 372, all of which have annular shapes aligned parallel to the longitudinal axis LA. The electrical connector 25A is configured to engage and couple with the electrical connector 25B. In particular, when the cartomizer 30 is connected to the control unit 20, the inner electrode 375 contacts the electrical contacts 250 of the control unit 20 to provide a first electrical path between the cartomizer and the control unit, and the outer electrode 370 contacts the body connector 240 of the control unit 20 to provide a second electrical path between the cartomizer and the control unit. Thus, the inner electrode 375 and the outer electrode 370 function as positive and negative terminals (or vice versa) for the cartomizer 30 to receive power from the battery 210 of the control unit 20.

[0033] The cartomizer 30 further includes a wick 362 and a heater 365. The wick 362 may be made of any suitable porous material, such as cotton, fiberglass, ceramic, or the like, and may extend from the reservoir 360 across and through the air passage 355. Similarly, the heater may be implemented in any suitable manner, for example, as a resistive heater in the form of a wire coil or metal mesh, a ceramic plate or disk, or the like. The heater 365 is electrically connected to terminals 370 and 375 via supply lines 366 and 367 to receive power from the control unit 20 (and the battery of the control unit 20). The wick 362 is located near the heater. For example, the heater may surround or be surrounded by the wick. Thus, liquid transported from the reservoir 360 by the wick 362 is heated by the heater 365 in response to a user inhaling on the electronic aerosol delivery device 20, generating vapor that flows along the air passage 355 and exits the mouthpiece 35.

[0034] It should be noted that various components and details have been omitted from Figures 2 and 3 for clarity. For example, detailed wiring, such as between connector 25B, circuit board 202, and battery 210, is generally not shown, as is the wiring between power supply lines 366, 367, and contacts 25A. Similarly, input / output features of system 10 (such as buttons or LEDs) are not shown.

[0035] It will also be understood that the configuration of the electronic aerosol delivery device 20 shown in FIGS. 1-3 is merely an example intended to provide an illustrative context for the present application. Those skilled in the art will recognize many potential variations. For example, rather than a two-part system (control unit 20 and cartomizer / cartridge 30), system 10 may be formed as an integrated device or alternatively may be formed from three or more sections. The aerosol-generating material may not be a liquid, but may include a solid, potentially in leaf or powder form (or gel or paste, etc.), as described above. In some implementations, the system may first generate a stream of heated vapor (e.g., steam), which passes through the aerosol-generating material, thus heating it and generating the aerosol. In some implementations, system 10 may include multiple different aerosol-generating materials and support a combination or selection of such materials. Some devices may include a removable cartridge containing reservoir 360, but the atomizer (e.g., heater 365) may not be included in the cartridge (e.g., the atomizer may be in a separate component). Additionally, rather than extending distally from the cartomizer to provide a linear airflow along the axis LA, the control unit 20 may have a folded configuration in other implementations. Furthermore, the heater 365 may be implemented in various forms, such as a planar mesh or ceramic heater. In some cases, the atomizer may be provided as some form of nebulizer (e.g., based on vibration rather than heating). Those skilled in the art will appreciate that these examples are only a few of the possible variations in the configuration of the electronic aerosol delivery systems disclosed herein.

[0036] FIG. 4 is a schematic diagram of certain electrical (including electronic) components of the control unit (aerosol delivery device) 20 of FIG. 1 . Note that at least some of these components are shown by way of example only and may be omitted (and / or supplemented or replaced with other components) depending on the circumstances of any given implementation. Furthermore, while the components shown in FIG. 4 are assumed to be located in the control unit 20 rather than the cartomizer 30 (because a given control unit can be reused with many different cartomizers 30), other configurations may be employed as desired. Additionally, while the components shown in FIG. 4 may be located on a single circuit board 202, other configurations may be employed as desired. For example, the components may be distributed across multiple circuit boards or may not even be mounted on a single circuit board. Furthermore, for clarity, FIG. 4 omits various elements typically present in this type of device, such as most power supply lines, memory (RAM), and / or (non-volatile) storage (ROM).

[0037] 4 includes the (rechargeable) battery 210 and connector 25B for coupling to the cartomizer (cartridge) 30, as previously described, as well as the (micro)controller 455, as described below. The battery 210 is further coupled to a USB connector 425, e.g., a micro, mini, or Type-C connector, that can be used to recharge the battery 210 from an external power source (typically via some rechargeable circuitry not shown in FIG. 4). Note that the battery 210 may support other forms of charging. For example, it may support charging via some other form of connector, wireless charging (e.g., inductive), charging via connector 25B, and / or charging by removing the battery 210 from the e-cigarette 10.

[0038] The device of FIG. 4 further includes a communication interface 410 that can be used for wired and / or wireless communication with one or more external systems (not shown in FIG. 4), such as a smartphone, a laptop, and / or other types of computers and / or other appliances. Wireless communication can be performed using (for example) Bluetooth® and / or any other suitable wireless communication standard. It will be understood that the USB interface 425 can also be used to provide a wired communication link instead of (or in addition to) the communication interface 410. For example, the USB interface 425 can be used to provide wired communication to the system, while the communication interface 410 can be used to provide wireless communication to the system.

[0039] Communications to and / or from electronic aerosol delivery device 20 can be used for a variety of purposes, such as to collect and report (upload) operational data from system 10 regarding usage levels, settings, any error conditions, and / or to download updated control programs, configuration data, etc. Such communications can also be used to support interaction between electronic aerosol delivery device 20 and external systems, such as a smartphone owned by a user of electronic aerosol delivery device 20. This interaction can potentially support a variety of apps, including collaborative or social media-based applications (apps).

[0040] The device of FIG. 4 further includes a motion sensor 465 (described below) and an airflow sensor 462 that detects when a user inhales on the system 10. Such detection can cause the microcontroller 455 to provide power from the battery 210 to the cartomizer 30 (particularly the heater 365) to generate a vapor output for the user to inhale (a process commonly referred to as puff actuation). The sensor 462 can detect airflow through any suitable mechanism, such as by monitoring changes in air flow and / or pressure. Note that some systems 10 do not support puff actuation. These systems are typically activated by the user pressing a button (or some other form of direct input). The microcontroller 455 can specify (and implement) one or more heating profiles for use with the heater 365, which determine the change in power level supplied to the heater 365 over time. For example, the microcontroller may provide most of the power from the battery 210 to the heater 365 at the beginning of a puff to quickly heat the heater 365 to an operating temperature, and thereafter the microcontroller may provide a reduced level of power to the heater 365 sufficient to maintain this operating temperature.

[0041] The device of FIG. 4 may further include user I / O functionality 420 supporting direct user input to system 10 (this user input / output may be provided instead of, or more generally, in addition to, the communication functionality described above). User output may be provided as one or more of visual, audio, and / or tactile output (feedback). For example, visual output may be achieved by one or more light emitting diodes (LEDs) or any other form of illumination, and / or a screen or other display, such as a liquid crystal display (LCD), capable of providing more complex forms of output. User input may be provided by any suitable functionality, such as by providing system 10 with one or more buttons or switches and / or a touchscreen (supporting both user input and output). As described below, user input may also be performed by movement of device 20 (or system 10 as a whole). Such movement is detected using motion sensor 465. In this case, motion sensor 465 may be considered part of user input / output functionality 420.

[0042] The microcontroller 455 may be located on the PCB 202. The PCB 202 may also be used to mount other components, such as the motion sensor 465 and / or the communication interface 410, as needed. Some components may be mounted remotely. For example, the airflow sensor 462 may be located adjacent to the airflow path through the system 10, and user input features (e.g., buttons) may be located on the external housing of the system 10. The microcontroller 455 generally includes a processor (or other processing functionality) and memory (ROM and / or RAM). The operation of the microcontroller 455 (and some other electronic components) is typically controlled, at least in part, by software programs running in the processor of the microcontroller 455 (or other electronic components, as needed). Such software programs may be stored in non-volatile memory, which may be integrated into the microcontroller 455 itself or provided as separate components (e.g., on the PCB 202). The processor may access the ROM or any other suitable store to load individual software programs for execution as and when needed. Microcontroller 455 may include appropriate interfaces (and control software) for interacting with other components of system 10 (such as those shown in FIG. 4 ). In addition, microcontroller 455 supports filter 470 and artificial intelligence (AI) system 480, shown generally in FIG. 4 and described in more detail below.

[0043] 4 may be modified by those skilled in the art. For example, the functionality of (micro)controller 455 may be distributed across one or more components, which in combination operate as a microcontroller. Furthermore, in some examples, filter 470 may be supported by motion sensor 465 rather than microcontroller 455 (depending on the capabilities of motion sensor 465). Alternatively, in some examples (not shown), filter 470 and / or AI system 480 may be supported by an external computing device configured to communicate with microcontroller 455 (e.g., via interface 410).

[0044] Additionally, there may be a PCB or the like provided in combination with battery 210 to control the recharging of the battery, such as detecting and preventing voltage or current overloads and / or excessively long charging times, as well as to control the discharging of the battery (e.g., to prevent the battery from being excessively discharged to the point of being damaged). It will be appreciated that the above set of alternatives and variations on the configurations of Figures 1-4 is by no means exhaustive, and that many more alternatives and variations will be apparent to those skilled in the art.

[0045] The motion sensor 465 provides sensitivity to motion of the system 10. In some examples, the motion sensor 465 is provided by an accelerometer or a gyroscope. In some examples, the motion sensor 465 is provided by a device, module, or unit that provides multiple types of movement-sensitive functionality. For example, the motion sensor 465 can be provided by or otherwise combined with an accelerometer and a gyroscope, and any other motion-sensitive components of the system 10. Such a device, module, or unit can be referred to as an inertial motion unit (or alternatively, an inertial measurement unit) instead of a motion sensor.

[0046] In some implementations, the motion sensor 465 is provided by a module available from STMicroelectronics, the LSM6DSLTR, used as a combined accelerometer and gyroscope (effectively a 2-in-1 system-in-package chip). In particular, this device provides a 3D digital gyroscope and a 3D digital accelerometer, i.e., three-axis sensitivity for both rotational and linear motion. Further details about this module can be found at https: / / www.st.com / content / st_com / en / products / mems-and-sensors / inemo-inertial-modules / lsm6dsl.html.

[0047] Note that the power consumption of the LSM6DSLTR device is on the order of 0.5 mA in an "always on" configuration. Assuming a typical capacity of 500 mA-hours for the battery 210, the power consumption of the motion sensor 465 per day represents 2.4% of the battery's capacity. Given that e-cigarettes are often recharged on a daily basis (vaporizing liquid typically requires relatively high current levels), this level of power consumption for the motion sensor 465 can easily be supported.

[0048] In some implementations, the microcontroller 455 is provided by an STM32F429ZIT6 module, available from STMicroelectronics, which includes an ARM Cortex-M4 core along with a digital signal processor, floating-point unit, and flash memory. This module includes a timer for pulse width modulation (PWM), typically used in e-cigarettes to vary the output from the heater 365 according to, for example, the heating profile described above. In particular, the PWM duty cycle can be decreased to reduce the amount of power delivered to the heater or increased to increase the power level. Further details are available at https: / / www.st.com / content / st_com / en / products / microcontrollers-microprocessors / stm32-32-bit-arm-cortex-mcus / stm32-high-performance-mcus / stm32f4-series / stm32f429-439 / stm32f429zi.html.

[0049] As described herein, the motion sensor 465, filter 470, and AI system 480 are used in combination to identify different user inputs or gestures to the electronic aerosol delivery device 20. This combination enables an efficient approach to gesture recognition. As described herein, by embedding the motion sensor 465 within the electronic aerosol delivery device 20, for example, on the circuit board of such device, the AI ​​system 480 can be trained and deployed to recognize consumer gestures (based on the motion data from the motion sensor 465) to complement or even completely replace mechanical movements. Advantageously, using the filter 470 to generate features from the motion data can significantly reduce the amount of data processed by the AI ​​system, thereby speeding up processing and gesture identification by the AI ​​system.

[0050] 5 shows a high-level overview of the process of identifying a user gesture, in which motion sensor 465 generates data samples 466 (e.g., motion sensor measurements or readings) representing the spatial motion of electronic aerosol delivery device 20 that are passed to filter 470. The data samples regarding the motion of device 10 provide a time series indicative of at least one of the position, velocity, and / or acceleration of device 10.

[0051] Data samples 466 may be passed to filter 470 as each measurement is recorded by motion sensor 465, as part of a batch of data samples after a set period of time or number of measurements, or in response to a stimulus such as a signal to the motion sensor (e.g., in response to a user pressing or releasing a button). It will therefore be appreciated that motion sensor 465 may include memory and processing capabilities to enable the storage and transmission of data samples 466.

[0052] Data samples 466 can be passed to filter 470 by providing them as a direct input to filter 470 or by placing data samples 466 in a memory accessible or readable by filter 470. For example, the memory can be memory of filter 470 or memory of controller 455 that is accessible to both filter 470 and motion sensor 465 (e.g., allowing for the storage of data and the reading of stored data).

[0053] In some examples, a filter is included as part of the output of the motion sensor, as part of the input of the AI ​​system, and / or as an intermediate component between the motion sensor and the AI ​​system. For example, filter 470 may be implemented by a hardware component or a software element of microcontroller 455. Thus, filter 470 may be supported by controller 455 of electronic aerosol delivery device 20. In some examples, filter 470 may be implemented by a hardware component or a software element (not shown) of motion sensor 465. Thus, filter 470 may be supported by motion sensor 460. In some examples, filter 470 may be implemented by a hardware component or a software element of an external computing device (e.g., a smartphone, not shown) that communicates with motion sensor 465 via communication interface 410. Thus, filter 470 may be supported by the external computing device.

[0054] Based on the received data samples 466, the filter 470 extracts (e.g., determines or calculates) one or more features 471 (i.e., one or more values ​​of statistical and / or mathematical properties of the received data samples), which are then passed to the AI ​​system 480. Thus, the AI ​​system 480 can be thought of as receiving the data samples in the form of extracted features. As described below, each feature 471 provides a value related to or otherwise dependent on the spatial motion of the electronic aerosol delivery device 20. The filter 470 can operate in the time and / or frequency domains. By filter, we mean a component configured to extract one or more features (e.g., values ​​of statistical and / or mathematical properties) from data samples stored in memory (e.g., stored in a table as illustrated by FIG. 6).

[0055] A feature may be a singular value or a collection of related values. For example, a feature may be (or be derived from) a mean, median, maximum, minimum, or correlation associated with all or a portion of the data samples stored in memory. It will be understood that mean, median, maximum, minimum, and correlation are well-known statistical and / or mathematical properties. A feature may be derived from a portion of the data samples 466. For example, a calculation (e.g., an average) may be performed on data samples 466 corresponding to one degree of freedom. Alternatively, a calculation may be performed on data samples 466 corresponding to one degree of freedom during a first timer period (e.g., the first half of the measured window), while further calculations may be performed on the same degree of freedom but over a different time period (e.g., a subsequent or overlapping time period). In some examples, a feature may be calculated based on a portion of the data samples corresponding to multiple degrees of freedom. For example, a value for the total movement of the device 20 may be calculated based on the degrees of freedom for the x, y, and z motion of the device.

[0056] Additionally, in some examples, extracted features can be used to influence the values ​​of other extracted features. For example, the latter features can be derived using techniques such as dynamic time warping to account for changes in gesture execution. Using dynamic time warping may require initial feature extraction to determine the speed at which the user is performing the gesture. By using dynamic time warping, the speed at which the user is performing the gesture can be adjusted when determining the value of the features.

[0057] It will be appreciated that the extracted feature(s) of data sample 466 are reduced in size (e.g., memory storage requirements) compared to data sample 466 extracted by filter 470. The number of features determined based on the data samples may be selected based on various operational parameters, i.e., the computing power available on the target computing device, the similarity of the gestures performed, and the number of gestures. In some examples, the filter is configured to extract the same features for each set of data samples. However, in some examples, the filter may be configured to extract only certain features depending on the data sample. For example, if the number of data samples is below a certain number, certain features may not be calculated if those features are deemed not useful for the reduced sample size.

[0058] AI system 480 uses the extracted features to identify or predict gestures 481 (e.g., user inputs) related to the spatial movement of electronic aerosol delivery device 20 represented by data sample 466. Thus, AI system 480 can be considered to operate as a classifier in that a particular feature (e.g., one or more values ​​of statistical and / or mathematical properties of the data sample) is considered to represent, indicate, or be associated with a particular gesture among a set of gestures (i.e., movement patterns corresponding to user inputs) known to AI system 480. AI system 480 outputs the identified gestures, thereby enabling further action (e.g., further processing) to be performed by control software of the electronic aerosol delivery system (or another electronic device) as needed based on the identified particular gesture. It will be appreciated that in some instances, AI system 480 may determine that the extracted features do not adequately correspond to any gesture in the set of gestures (e.g., the relationship or correlation between the extracted features and each individual gesture may be below a threshold). For example, it will be appreciated that the electronic aerosol delivery system may move when carried, for example, in a pocket or bag. Such movements are not intended to cause the gesture to be recognized by the AI ​​system 480.

[0059] FIG. 6 illustrates an example data structure for storing data samples 466 corresponding to measurements or readings obtained by the sensing component of the motion sensor 465, where the data structure is in the form of a table 610. The data samples 466 record or detail the spatial motion of the electronic aerosol delivery device 20. It will be appreciated that there are various alternative data structures (e.g., matrices) known to those skilled in the art that can be used to store the data samples. In some examples, the table 610 and associated data samples 466 can be stored in a buffer maintained in memory of the electronic aerosol delivery system (e.g., memory associated with the microcontroller 455). Alternatively, or additionally, the table and associated data samples can be stored in a buffer maintained in memory of an external device that communicates with the electronic aerosol delivery system (in some examples, the external device can include a copy of the data stored in the electronic aerosol delivery system). The memory can be the memory of the filter 470 or the memory of the controller 455 and is accessible to both the filter 470 and the motion sensor 465 (e.g., allowing for the storage and / or reading of data).

[0060] The illustrated table has a window size or length 620 (i.e., number of data rows) corresponding to the number of data readings, frames, or samples stored in the table, and a width 630 (i.e., number of data columns) matching the number of axes or degrees of freedom being sampled (e.g., a 3D accelerometer has a width of 3, while a motion sensor combining a 3D accelerometer and a 3D gyroscope has a width of 6). The window size 620 indicates the number of samples to examine from the motion sensor to predict a gesture, and the width 630 indicates the number of degrees of freedom for predicting a gesture. As shown, the table of FIG. 6 has at least three columns, with the three illustrated columns corresponding to different axes (x, y, z) of the gyroscope. Additionally, the table of FIG. 6 has a window size 620 of at least 3. Each entry 650 in the table corresponds to a data sample 466 for a particular sampled degree of freedom. The specific nature and format of the data samples 466 passed from the motion sensor 465 to the filter 470 will depend on the particular implementation and situation. For example, in the case of a six-axis motion sensor, the first set of columns of data samples 466 in the table will contain values ​​representing linear acceleration (e.g.,

number

number

[0061] In some examples, the window size 620 is fixed. For example, the window size 620 may be fixed to correspond to the number of measurements (i.e., number of samples) taken within the duration of a gesture. For example, if it takes a user up to 3 seconds to perform a gesture, the window size 620 may correspond to 3 seconds multiplied by the sampling rate (i.e., the number of measurements per second). For example, if the motion sensor 465 has a sampling rate of 1 kHz, the window size 620 would be 3000. It will be understood that the window size 620 depends on the sampling rate and the duration of the measurement period. The sampling rate of the motion data and the duration of the measurement period are sufficient to provide an accurate indication of the device's motion.

[0062] It will be appreciated that in some examples, different aspects of the motion sensor 465 may have different sampling rates. For example, the gyroscope of the motion sensor 465 may have a sampling rate of 0.5 kHz, while the accelerometer of the motion sensor 465 may have a sampling rate of 1 kHz (or vice versa). Thus, for a table 610 having a particular duration, the number of data samples in the column 630 for the gyroscope axis will be half the number of data samples in the column for the accelerometer axis. In other words, it is not necessary for each axis of the table to have an equal number of rows 620 (or, if the number of rows is equal, some rows in the column corresponding to measurement devices with lower sampling rates will be empty).

[0063] In some examples, a new data sample (for each axis) is added every 1 / (sampling rate) seconds. In some examples, table 610 may be initially empty, and new data samples may be added to table 610 incrementally. In some examples, after a column of table 610 is filled, new data samples are not added to table 610. In these examples, the contents of table 610 are used to extract features when the number of entries 630 in a column equals window size 620. For example, as data samples 466 fill the table, filter 470 extracts features and provides the extracted features to AI system 480 for analysis by AI system 480 to determine the corresponding gesture performed by the user (or that a recognized gesture was not performed).

[0064] In some examples, new data samples may be added incrementally to a table column to replace data samples included in the table 610 column after the table column is filled. For example, new data samples replace the oldest data sample in the table column (i.e., a moving window). In these examples, new data samples may be added (the AI ​​system is continuously or periodically provided with data samples) until the AI ​​system 480 identifies a gesture from features extracted from the data samples in table 610. In effect, this causes the window represented by data samples 466 to slide by a fraction of the window's duration. New data samples 480 may then be added iteratively (either singly or in batches), with the filter continually extracting features and the AI ​​system 480 re-evaluating the features after each addition. In this way, the filter can be thought of as extracting features at a particular extraction or data rate. This rate can be referred to as the amount of data generated per unit time and depends on how often features are computed and the size of the features in memory. Because multiple data samples are typically measured per feature computation, this rate is typically much smaller than the sampling rate. For example, if data samples recorded at 1 kHz and spanning 1 second are used to create a single feature having a size equal to a single data sample per second, the feature extraction rate is 1000 times smaller than the data sampling rate.

[0065] In some examples, new data samples can be added until a determination is made that the electronic aerosol delivery device 10 has not been moving for a period of time. For example, in some implementations, the motion sensor 465 can distinguish between periods of movement and periods of no movement, the latter being determined when there is little or no change in the values ​​of the data samples. In these implementations, the motion sensor 465 may have built-in functionality (e.g., processing power) that enables the motion sensor 465 to perform analysis. In other examples, new data samples can be added to the table 610 until the user indicates that they have stopped the gesture (e.g., by pressing or ceasing to press a button).

[0066] The duration of the measurement period (i.e., the number of rows in the example table of FIG. 6 ) can be set based on a reasonable maximum duration it takes to perform any gesture in a sequence of gestures. For example, different gestures may take different durations or time periods to perform (e.g., tracing a circle with the electronic aerosol delivery device 20 may take longer than a user moving the electronic aerosol delivery device 20 upward in a vertical line). Therefore, to be able to identify a gesture based on stored data samples, the data samples must cover a duration that allows any gesture to be distinguished from any other gesture. As mentioned above, in some examples, this may require setting the duration of the measurement period based on a reasonable maximum duration it takes to perform any gesture in a sequence of gestures. Of course, in some examples, a particular gesture may be recognizable from measurements within a shorter time period than it takes to perform the entire gesture, if the particular gesture can be distinguished from measurements over a shorter time period (e.g., a user may trace a circle to provide user input, but the window size may include data corresponding to the spatial movement of only a portion of the arc of the circle, as long as the arc allows the circular gesture to be identified). In some examples, the duration of the measurement period may be less than 5 seconds, preferably less than 3 seconds, and more preferably less than 2 seconds.

[0067] In some examples, data samples 466 may be recorded in table 610 (e.g., incrementally added to the table) as long as a criterion is met. In some of these examples, window size 620 may vary such that the duration of the measurement period depends on the length of time the criterion is met. In some examples, the criterion may be that the user is interacting with the electronic aerosol delivery device to indicate that they are gesturing (e.g., by interacting with a user input feature, such as a button that is pressed while the user is gesturing). In some of these examples, the user may indicate both the beginning and end of a gesture. Thus, device 20 may be configured to allow the user to indicate the beginning and / or end of a gesture, for example, by pressing a button on device 20 or tapping device 20. In some examples, motion sensor 465 may respond to a signal to stop transmitting or to the detection of a period of no motion by ceasing transmission of data samples 466 from the motion sensor to filter 470 or memory accessible by the filter. Thus, the motion sensor may be turned off so that no data samples are generated. In some of these examples, table 610 may be emptied before new data samples are added.

[0068] In some implementations, the device can operate in several different states. A state of device 20 refers to a state or mode of operation (e.g., how the device is configured to operate) and may include a state or mode of operation of microcontroller 455 implementing filter 470, AI system 480, and / or other components of the device. For example, a first state may correspond to the device being in a dormant state (e.g., a low-power state), and a second state may correspond to the device being in an active state. Alternatively, the first state may operate when a user is not controlling or interacting with device 20 (e.g., via a user input function such as a button), and the second state may operate when a user is controlling or interacting with device 20. In some examples, the system may control when to use or operate an AI system to identify user input based on whether the device is in the first state or the second state (e.g., allowing a user to control a device on which an AI system is operating via a user input function that toggles the device between states). Additionally, when the AI ​​system 480 is not being used to distinguish between different user inputs, the motion sensors may either be turned off and no data samples may be generated, or data samples may be generated but not provided to the AI ​​system 480.

[0069] During this first state, the sample measurement rate (i.e., data rate) of the motion sensor 465 may be reduced or stopped, and the filter 470 may extract features less frequently, if at all (e.g., to conserve power). For example, features 471 may not be extracted by the filter 470 from data samples recorded while the device was in the first state (e.g., the filter may be deactivated). When the microcontroller 10 or device 20 switches to the second state, the filter may be activated and begin extracting features. In some examples, the filter may extract features based on data samples 466 that include measurements taken before the device switched to the second state, while in other examples, the filter 470 may extract features based on data samples 466 that consist of measurements taken only after the device switched to the second state. For example, table 610 (or other data structure) may be emptied between the first and second states, while in other instances table 610 is constantly updated with new values, such that new values ​​obtained while the device is in the second state eventually push out values ​​obtained while device 20 is in the first state (unless data sampling is interrupted).

[0070] In some examples, the filter 470 is not deactivated during the first state; instead, the filter is configured to extract a first set of one or more features from the data samples in the first state of the device 20, and to extract a second, different set of one or more features from the data samples in the second state of the device 20. In some of these examples, the sample rate of the motion sensor 465 and the rate at which the filter 470 extracts features 471 may be increased (e.g., the filter may extract features more frequently) when the device switches from the first state to the second state.

[0071] It will be understood, therefore, that in some examples, a first state is a state in which the AI ​​system is not used to distinguish between different user inputs (e.g., AI system 480, filter 470, or controller 455 implementing the AI ​​system are in an inactive mode), and a second state is a state in which the AI ​​system is used to distinguish between different user inputs (e.g., AI system 480, filter 470, or controller 455 implementing the AI ​​system are in an active mode). Thus, AI system 480, or controller 455 implementing AI system 480, can be switched or placed into an active mode that enables gesture recognition when device 20 switches from the first state to the second state.

[0072] It will be appreciated that in some of these examples, the motion sensor 465 is configured to provide data samples regarding the system's motion directly to the AI ​​system 480, rather than the AI ​​system 480 receiving the data samples through a filter that extracts features from the data samples. In other words, in some example systems, there may be no filter, and the AI ​​system receives unfiltered data samples. The AI ​​system 480 in these examples is configured to receive unfiltered data samples from the motion sensor (e.g., by the motion sensor inputting the unfiltered data samples into memory accessible to the AI ​​system) and use the unfiltered data samples to identify different user inputs to the system. As mentioned above, in some examples, the AI ​​system can employ multiple AI models to identify user inputs (i.e., gestures). Each AI model is trained or configured for a particular user input. Additionally, a user input function (e.g., a button or slide) can be provided that is configured to control when the AI ​​system is used to identify different user inputs. This can reduce the load on any processor associated with the AI ​​system 480 by only using the processor to employ the AI ​​system 480 to identify gestures when a user controls the device 20 using the user input function to identify gestures.

[0073] In some examples, device 20 is configured to switch states in response to a user interacting with a user input feature (e.g., a button). Thus, in these examples, the user uses the user input feature to control when the AI ​​system is used to distinguish between different user inputs (i.e., gestures). This can be used to allow exercise to be processed only when desired by the user, which can prevent accidental activation and also reduce energy consumption. In other examples, device 20 is configured to switch states in response to a determination that device 20 has been moved. In these latter examples, the device may be configured, in a first state, to detect or otherwise determine whether device 20 has moved or is being moved. Similarly, this can be used to reduce energy consumption by only enabling certain processing features after a period of exercise.

[0074] FIG. 7 illustrates an example data structure containing feature values, where the example data structure is in the form of a table 710. It will be appreciated that there are a variety of alternative data structures (e.g., matrices) known to those skilled in the art that can be used to store data samples. As previously discussed, features 730 (each feature having a corresponding column as shown in FIG. 7 ) can be a value 720 or a collection of values. For example, a feature 740 among features 730 can be (or be derived from) a mean, median, maximum, minimum, or correlation associated with all or a portion of data samples 466 stored in memory accessible by filter 470. Furthermore, features 730 can be derived or calculated using techniques such as dynamic time warping to account for variations in gesture execution. It will be appreciated that the extracted feature(s) 730 of data sample 466 have reduced size (e.g., memory storage requirements) compared to data sample 466 extracted by filter 470. In other words, the memory required to store a value (i.e., feature 740) such as the mean of a numerical sequence is significantly less than the memory required to store the numerical sequence. Therefore, given that the extracted features 740 are generated less frequently than the data samples provided by the motion sensor, the extracted features 740 can typically be considered to have a lower data rate than the data rate of the data samples.

[0075] The number of features 730 determined or extracted based on the data samples can be selected based on various operational parameters associated with the electronic aerosol delivery device 20 and / or any individual devices communicating with the aerosol delivery device 20 via the communication interface 410. For example, this selection can be based on factors or parameters such as available computing power, the similarity of the gestures performed, the number of gestures, and the number of measured degrees of freedom. For example, in systems with greater computing power, the number of computed features may be greater, potentially improving the reliability and accuracy of gesture identification. In some exemplary systems where there are two or more similar gestures that can be performed, more features may be required to accurately distinguish between these features. Similarly, the more gestures there are, the more features may be required to accurately distinguish between these features compared to systems configured to recognize fewer gestures. Additionally, the presence of more sophisticated measurement devices offering more measured degrees of freedom allows for the measurement of more features, thereby improving accuracy and reliability.

[0076] FIG. 8 illustrates an exemplary data structure in the form of a lookup table 810 associated with the AI ​​system 480 and containing values ​​for use in identifying gestures associated with the extracted features 730. For example, the value 840 may be a reference value for comparison with the extracted features 730. As shown, the table 810 may include several rows corresponding to different gestures 820. Each row includes a set of values ​​840 corresponding to the respective gesture. As further shown in the figure, the table 810 may include several columns 830 corresponding to different features, with each gesture 820 having a value 840 (i.e., feature) for each column 830. The values ​​for each gesture stored in the lookup table 810 are used by the AI ​​system 480 in determining or identifying the respective gesture corresponding to the features 730 extracted by the filter 470. Each row 820 of the lookup table 810 may be considered to represent a known or pre-programmed gesture, in that a value corresponding to each gesture is stored in the lookup table 810 before the user moves the electronic aerosol delivery device 20. The data 840 stored in the lookup table 810 can be thought of as training data for training the AI ​​system 480.

[0077] In some examples, the values ​​stored in lookup table 810 are obtained using a motion sensor and filter configured according to motion sensor 465 and filter 470 of electronic aerosol delivery device 20. In some examples, the filter is provided by a separate computing platform (e.g., a platform with greater processing power than filter 470 of electronic aerosol delivery device 20). To acquire data, a test electronic aerosol delivery system (e.g., part of a data logging system including electronic aerosol delivery device 20 and another computing platform) can be moved according to a gesture trained into AI system 480 (i.e., performing the gesture). The motion sensor of the test electronic aerosol delivery system can be used to measure motion (e.g., the motion sensor can generate data at a fixed frequency or sampling rate). Data corresponding to the measured motion is provided to the filter (e.g., via a communication interface if the filter is external, or via internal circuitry if the filter is part of the test electronic aerosol delivery system). The filter is then used to extract features based on data corresponding to the spatial motion of the test device (i.e., the measured motion).

[0078] In some examples, the data 840 in the lookup table 810 corresponds to features extracted by the test system after a single execution of the gesture. In other examples, a movement corresponding to a target gesture is executed multiple times (e.g., at least three separate executions), each time measuring the movement and extracting a respective feature. In these examples, the value may correspond to an average of the features extracted by the test system in each separate execution of the same target gesture.

[0079] It will be appreciated that the filter used to generate the training data may differ from the filter 470 of the electronic aerosol delivery device 20 in that it may be configured to generate more feature 830 values. The number of extracted features 830 may be greater, at least because it is not necessary to provide a rapid response to the user. Conversely, the filter 470 of the electronic aerosol delivery device 20 may be configured not to extract unnecessary features 471 (i.e., features that do not significantly affect gesture selection), because the AI ​​system 480 provided with the training data can identify features necessary for gesture identification. Reducing the number of features to only those that need to be extracted and processed to identify relevant gestures can improve the response time of the AI ​​system 480.

[0080] To provide accurate training data, each individual execution of each gesture is separated from movement data that occurs outside of the gesture execution. For example, if there is 10 seconds of movement data and the execution occurs between 2 and 5 seconds, the movement data from 2 seconds before and 5 seconds after can be discarded to separate the individual executions of the gesture.

[0081] In some examples, separation can be achieved by comparing movement data in any available measured degrees of freedom with a movement threshold, thereby allowing the beginning and end of execution to be determined. For example, movement recorded by an accelerometer and / or gyroscope can be used to separate gesture execution. Furthermore, in some examples, movement recorded by any possible combination of accelerometer and / or gyroscope degrees of freedom is used to separate gesture execution.

[0082] In some examples, the separation can be achieved depending on a condition (or conditions) that indicates the duration of the gesture. In other words, the condition is used to set or otherwise determine the start and end of execution. In some examples, a button (or other actuator) may be pressed for the duration of the gesture's execution, or the button may be pressed at the beginning and end of execution. In some examples, a light (e.g., from an LED) or a vibration (e.g., from a haptic device) indicates the start and end of recording of individual gestures. By separating gestures, filters can be used to extract features only from data related to the gesture, rather than data associated with periods of non-movement or other movement that could otherwise skew feature extraction. Thus, features extracted from the separated gestures are more accurately related to or representative of each gesture.

[0083] In some examples, the data in lookup table 810 includes optimal or representative values ​​of features for known gestures that a user may perform. The optimal or representative values ​​are obtained by extracting features from motion data for one or more executions of each respective gesture. In some examples, the optimal or representative values ​​correspond to features extracted by the test system from data corresponding to individual executions of a target gesture. In some examples, the optimal or representative values ​​correspond to an average of each feature extracted by the test system from data corresponding to multiple individual executions of the same target gesture.

[0084] In these examples, the lookup table 810 can be used by the AI ​​system 480 to directly compare features 471 extracted by the filter 470 during use of the electronic aerosol delivery device 20. Direct comparison means that the AI ​​system 480 compares the values ​​in the lookup table with features extracted by the filter 470 (e.g., features stored in a data structure according to FIG. 7 ) to determine the most likely match from known gestures. For example, the AI ​​system 480 can determine which set of features corresponding to known gestures in the lookup table are most similar to the extracted features 471. It will be appreciated that the AI ​​system 480 can be configured to implement a wide range of statistical and computational structures, such as neural networks, support vector machines, Bayesian classifiers, and machine learning systems, to establish which known gestures (e.g., trained gestures) are most similar to the extracted features 471. In this manner, the AI ​​system 480 can identify or predict which of the trained gestures the user intends to input.

[0085] In some examples, AI system 480 can determine or identify weights or biases associated with particular features. For example, AI system 480 can identify that an exact match between the values ​​in lookup table 810 and the values ​​of a particular extracted feature indicates a high probability of a particular gesture (i.e., user input). Similarly, AI system 480 can identify that a combination of two of more extracted features having similar values ​​to the respective values ​​stored in the lookup table (e.g., the ratios implied in the combination are similar for the extracted features and the stored values) can indicate a high probability of a particular gesture. It will be appreciated that by providing training data to AI system 480, AI system 480 can identify unique correlations between feature values ​​and gesture execution based on the training data, which can be used to determine or identify which gesture is being performed by a user of electronic aerosol delivery device 20.

[0086] In some examples, the data in the lookup table 810 corresponds to values ​​used in generating an AI model to which the extracted features are passed. For example, each row in the lookup table may comprise a set of values ​​used as inputs for a respective AI model implemented by the AI ​​system. The AI ​​system may comprise multiple AI models, each corresponding to a different user input (i.e., gesture) to the electronic aerosol delivery device. By corresponding, we mean that each AI model is trained to identify a respective one of the different user inputs (i.e., known gestures) to the electronic aerosol delivery device.

[0087] In these examples, the AI ​​system is configured to provide the extracted features to each of the AI ​​models to identify different respective user inputs to the device. For example, each of the AI ​​models can provide an output indicating the likelihood or probability that the extracted features correspond to the respective AI model. The AI ​​system 480 can then identify that the user performed the user input or gesture indicated as most likely. In some cases, the AI ​​system 480 may be configured to identify that the user did not provide the correct gesture, for example, if the output of each model indicates a probability below a threshold. In some examples, different thresholds can be used for the outputs from different models. As discussed above, it will be appreciated that the electronic aerosol delivery system may move when carried, for example, in a pocket or bag. Implementing a threshold reduces the likelihood that such movement will cause a gesture to be recognized by the AI ​​system 480.

[0088] AI models for implementation by AI system 480 can include a wide range of statistical and computational structures, such as neural networks, support vector machines, Bayesian classifiers, machine learning systems, etc. For example, in some implementations, AI model 480 is provided using the TensorFlow Lite platform, originally developed by Google and later released as an open-source deep learning framework for on-device inference (see https: / / www.tensorflow.org / lite). Alternative platforms that can be used for AI model 480 include PyTorch, originally developed by Facebook® and later released as an open-source machine learning library (see https: / / pytorch.org / ), and / or the Microsoft Cognitive Toolkit (CNTK), an open-source toolkit for distributed deep learning (see https: / / docs.microsoft.com / en-us / cognitive-toolkit / ).

[0089] The gestures trained into the AI ​​system 480 for recognition are symbolic in nature. In other words, the user's device movements are not used to provide a direct analog of some physical parameter, such as position or velocity (as might be used, for example, in a gaming context). Rather, user input is used to perform a selection (classification) from a discrete (finite) set of different possibilities. Each possibility is associated with a respective gesture. The AI ​​system 480 is trained to match specific examples of movement data 466 generated by user input (device movements) to corresponding gestures. It will be appreciated that the training phase 910 can be iterative in nature. For example, if the AI ​​system 480 has difficulty distinguishing between two different gestures, it may modify the sequence of gestures to modify or eliminate one of the problematic gestures (or potentially replace the eliminated gesture with another, more recognizable gesture).

[0090] More generally, system 10 may have various levels of configurability. For example, in some implementations, AI model 480 may be finalized (fixed) and unchangeable. In this case, a user may be provided with a set of gestures to use, as in a hardcopy instruction manual.

[0091] In some implementations, updates may be provided by the supplier of electronic aerosol delivery device 20 to update the functionality of AI system 480. In some implementations, a user may be able to replace or supplement the data used to train AI system 480 (and any AI models) by performing gestures (e.g., as part of a calibration phase). Filter 470 may be used to extract features, which are used to provide values ​​used in identifying gestures associated with the extracted features. The updated values ​​may be personalized to the user in that they take into account how the user performs each gesture, thereby allowing for more accurate identification of the user's future gestures.

[0092] In some implementations, a user may be able to supplement existing gestures by providing additional training data for recognizing new gestures. In some of these cases, AI system 480 may implement a new AI model for each new gesture. Thus, a user may be able to add a newly created gesture to an existing set of gestures or, in some cases, create an entirely new set of gestures (which may be personalized for the user). Updates (e.g., by training new gestures) can be used to provide at least one of the following: enhanced recognition, support for additional modes of operation, and support for additional input symbols for the local language.

[0093] Depending on the capabilities of system 10, such modifications to AI model 480 may be performed on an external device, such as a smartphone or laptop, using communications interface 410. In such cases, the external device can obtain the existing AI system 480 and filter 470 (whether from electronic aerosol delivery device 20 or some other suitable source), update the system, and reload the AI ​​system, and optionally the filter, back into electronic aerosol delivery device 20. It will be appreciated that the external device may have significantly improved processing power, such that extraction of features from new motion data and training of the system based on those features can be improved by utilizing the external device. Software may be provided or made available by the supplier of electronic aerosol delivery device 20 to facilitate such updates to AI system 480 and to guide the user.

[0094] As previously mentioned, filter 470 and AI system 480 typically operate on electronic aerosol delivery device 20. For example, as shown in FIG. 4, filter 470 and AI system 480 can be implemented on microcontroller 455, which effectively acts as the computing device of electronic aerosol delivery device 20 for operating filter 470 and AI system 480.

[0095] However, it is also possible for the electronic aerosol delivery device 20 to interact with some external device, such as a smartphone or laptop (e.g., using the communication interface 410), and offload some or all of the processing related to identifying the gesture to the external device (in which case the identified gesture 481 may be returned to the user via the external device, such as using a laptop screen, in addition to or instead of being returned to the user via the electronic aerosol delivery device 20).

[0096] For example, filter and / or AI system 480 may be located in an external system, such as a smartphone or laptop, using communication interface 410 (see FIG. 9 below). In these examples, electronic aerosol delivery device 20 can interact with the external device to offload some or all of the processing associated with filter and / or AI system 480 to the external device. The external system may have greater processing power such that the processing associated with filter and / or AI system 480 can be performed in a shorter time (e.g., because the external device complements the processing power of the electronic aerosol delivery system). Thus, offloading some or all of the processing can accelerate the response time for identifying and acting on a user's gestures.

[0097] In some examples, electronic aerosol delivery device 20 is configured such that when electronic aerosol delivery device 20 is in communication with an external device having filter 470 and / or AI system 480, electronic aerosol delivery device 20 offloads some or all of the processing associated with filter 470 and / or AI system 480 to the external device, but when electronic aerosol delivery device 20 is not in communication with an external device having AI system 480, electronic aerosol delivery device 20 performs all of the processing associated with filter 470 and / or AI system 480. This allows electronic aerosol delivery device 20 to advantageously utilize the processing capabilities of the external device when the external device is not in communication with electronic aerosol delivery device 20 without interfering with use of filter 470 and / or AI system 480.

[0098] Alternatively, in some examples, filter and / or AI system 480 is not installed in electronic aerosol delivery device 20, and thus electronic aerosol delivery device 20 cannot perform the processing associated with filter and / or AI system 480, but instead relies on an external device having filter and / or AI system 480 to perform the processing associated with filter and / or AI system 480. This has the advantage that the capabilities of microcontroller 455 (e.g., processor speed and memory requirements) can be reduced compared to a microcontroller 455 configured to support a filter and / or AI system.

[0099] Figure 9 provides a high-level schematic diagram of certain electrical components of the control unit and external device 910 of Figure 2. Figure 9 is generally similar to Figure 4, and therefore the operation of the individual components will not be described in detail. However, Figure 9 differs from Figure 4 in that instead of the AI ​​system 480 being implemented in the microcontroller 455, it is implemented in the external device 910, and the microcontroller 455 is configured to transfer extracted features to the AI ​​system 480 via the communication interface 410.

[0100] An external device is a device that is capable of performing any necessary processing related to identifying gestures (e.g., has memory and a processor capable of performing any required operations) using AI system 480. In some examples, external device 910 is a device such as a smartphone, a laptop, another type of computer, or another electronic appliance.

[0101] The microcontroller 455 is configured to communicate with the external device 910 via the communication interface 410. The exemplary communication interface 410 can provide wired and / or wireless communication. For example, the communication interface 410 can be configured to perform wireless communication using (for example) Bluetooth and / or any other suitable wireless communication standard. It will be understood that communication between the electronic aerosol delivery device 20 and the external device 910 is bidirectional in that the microcontroller 455 is configured to both send and receive data from the external device 910. For example, the microcontroller 455 can send extracted features 470 to the external device 910 with the AI ​​system 480 and can receive an indication of an identified gesture 481 from the external device 910 with the AI ​​system 480.

[0102] 9 relies on external device 910 having AI system 480 to perform processing associated with AI system 480. Accordingly, the capabilities (e.g., processor speed and memory requirements) of microcontroller 455 can be reduced compared to a microcontroller 455 configured to support an AI system. For example, given that microcontroller 455 is not used to implement AI system 480, the processor load is reduced.

[0103] Additionally, providing the microcontroller 455 with the filter 470 reduces the amount of data transmitted to the external device 910. For example, as previously described, the extracted feature(s) of the data sample 466 will be smaller in size (e.g., memory storage requirements) compared to the data sample 466 from which those features were extracted by the filter 470. Thus, by first processing the motion data 466 to extract features before transmitting the extracted features 471 to the AI ​​system 480, the amount of data transmitted over the communications interface 410 is significantly reduced. Thus, the extracted features can be considered to have a lower data rate than the data rate of the data samples provided by the motion sensor. This can be particularly advantageous when there is limited bandwidth available for transmission and / or when there are competing uses for the same transmission channel.

[0104] In some example devices 20, the AI-based input mechanism may interact with or be supported by other elements of the user interface to support the operation of the AI ​​system 480. For example, the system may be configured to prompt the user for user input in the form of a gesture. Similarly, the system may be configured to notify the user when a character input has been successfully identified. This prompt and / or confirmation may take a variety of forms, such as an audible output (e.g., a beep), haptic feedback utilizing vibrations of the system 10, and / or a visual output, such as provided by an LED lamp.

[0105] It will be appreciated that prompting / confirming user input, or the user indicating the beginning / end of such input, can facilitate the AI ​​model's recognition of input characters (such as by helping to determine the start of a recognition window, as in Figure 6). Similarly, feedback from the device to the user can help confirm that the user input was recognized as intended.

[0106] The use of motion sensors 465, filters 470, and AI system 480 to provide and recognize user input can be utilized in many different ways for system 10. The following examples are provided by way of illustration and not limitation. Any given device may support none, one, some, or all of these examples.

[0107] In some systems, user input recognized by the AI ​​system 480 may comprise a passcode or password, similar to a personal identification number (PIN), for enabling (authorizing) operation of the device 20 (and / or the complete system 10). For example, the passcode may comprise a series of gestures for recognition by the AI ​​system 480. If this passcode is not entered, some functionality of the device 20 or the entire system 10 may be locked or limited. For example, a heater may not be activated to prevent vaping. In some systems, user input recognized by the AI ​​system 480 may be used to set one or more operating parameters of the system 10. In some cases, this may involve entering both a gesture for the parameter's identifier and a gesture for the value of the operating parameter of interest. For example, the system 10 may support multiple heating levels during vaping, and the AI ​​system may be utilized to set a desired heating level, such as low, medium, or high. Other examples of user input to the AI ​​system may include resetting an error condition, selecting a desired heating profile, navigating a menu structure, controlling and performing data communications with an external device such as a smartphone, etc.

[0108] It will also be appreciated that using the motion sensor 465, filter 470, and AI system 480 as user inputs may help to complement (but not necessarily replace) existing user input functions. An example of this is a mechanical on / off button that physically opens or closes a circuit link (having a physical disconnect in the off state may, for example, provide slightly greater protection against accidental activation of the system).

[0109] More generally, the electronic aerosol delivery system disclosed herein incorporates a motion sensor configured to provide data samples related to the device's motion, a filter configured to extract features from the motion sensor data samples, and an AI system configured to receive the extracted features and use the extracted features to identify different user inputs to the device. As previously mentioned, the system may include an external, independent computing device, such as a smartphone or computer, that supports the AI ​​system (and in some instances, the filter). This system enables an efficient approach to gesture recognition, where feature extraction can reduce the data processed by the AI ​​system 480, thereby improving system responsiveness. Additionally, the amount of data transmitted between system components (including, in some instances, external devices) can also be reduced by feature extraction by the filter 470.

[0110] 10 is a schematic flowchart illustrating a process for identifying user input. The process is a method of operating an electronic aerosol delivery system, the system including a motion sensor configured to provide data samples related to the system's motion, a filter configured to extract features from the motion sensor data samples, and an artificial intelligence (AI) system. As noted above, in some examples, the motion sensor, filter, and AI system may be provided in electronic aerosol delivery device 20, but in some examples, the AI ​​system, and possibly the filter, may additionally or alternatively be supported by a separate external computing device.

[0111] The method begins at step 1010, in which a motion sensor provides data samples related to the motion of the system. The motion sensor operates as described above for a motion sensor that measures, records, or otherwise acquires data samples using one or more components configured to detect spatial motion of the device. For example, as described above, the motion sensor may include an accelerometer, a gyroscope, and other components for detecting spatial motion of the device. Each component for detecting spatial motion may be configured to detect and record motion along multiple axes (e.g., mutually perpendicular x-, y-, and z-axes). The motion sensor is configured to provide the data samples to the filter, for example, by inputting the data samples into a buffer accessible to the filter. If the filter is provided by an external computing device, the data samples may be provided (e.g., transmitted) via a communications interface.

[0112] The method continues at step 1020, where the filter extracts features from the motion sensor data samples. The extracted features include values ​​or relationships of statistical or mathematical properties of the motion sensor data samples, such as the mean, median, maximum, minimum, and correlation of a portion of the motion sensor data samples. The extracted features are provided to the AI ​​system, for example, by being stored in a memory accessible to the AI ​​system. When the filter is provided by the electronic aerosol delivery device and the AI ​​is provided by an external computing device, the features may be provided (e.g., transmitted) via a communications interface.

[0113] The method continues at step 1030, where the AI ​​system uses the extracted features to identify different user inputs to the system. The AI ​​system can input the extracted features into one or more AI models to identify or predict the user's desired input (e.g., by selecting the most probable input). The AI ​​system is thus configured to output the identified gestures for use by components of the system. For example, the identified user inputs can be used to control aspects of the system, such as an aerosol generator or a display. In this manner, the process shown in FIG. 10 provides an AI-supported user input function.

[0114] 11 is a schematic flow chart illustrating a process for identifying user input. The process is a method of operating an electronic aerosol delivery system, the system including a motion sensor configured to provide data samples related to the motion of the system, an artificial intelligence (AI) system configured to receive the data samples and use the data samples to identify different user inputs for the system, and a user input function configured to control when the AI ​​system is used to identify the different user inputs. As mentioned above, in some examples, the motion sensor and AI system may be provided in electronic aerosol delivery device 20, although in some examples, the AI ​​system is additionally or alternatively supported by a separate external computing device.

[0115] The method begins at step 1110, in which a motion sensor provides data samples related to the motion of the system. The motion sensor operates as described above for a motion sensor that measures, records, or otherwise acquires data samples using one or more components configured to detect spatial motion of the device. For example, as described above, the motion sensor may include an accelerometer, a gyroscope, and other components for detecting spatial motion of the device. Each component for detecting spatial motion may be configured to detect and record motion along multiple axes (e.g., mutually perpendicular x-, y-, and z-axes). The motion sensor is configured to provide the data samples to the AI ​​system, for example, by inputting the data samples into a buffer accessible to the AI ​​system. If the AI ​​system is instead provided by an external computing device, the data samples may be provided (e.g., transmitted) via a communications interface. Additionally, in some examples, the motion sensor provides the motion sensor data samples through a filter that filters the motion sensor data samples to extract features (the filter operates as described above (e.g., according to step 1020)). In these examples with filters, the data samples of the motion sensor can be thought of as a form of extracted features.

[0116] The method continues at step 1120, where the user controls when the AI ​​system is used to identify different user inputs via the user input feature. For example, when the user interacts with the user input feature to control the device to activate the AI ​​system (e.g., by pressing a button), the AI ​​system is enabled (e.g., activated) and is able to process motion sensor data samples to identify user inputs. Furthermore, when the user interacts with the user input feature to deactivate the AI ​​system (e.g., by ceasing to press a button), the AI ​​system is disabled and is no longer able to process motion sensor data samples to identify user inputs. Thus, by enabling the processor only when the user controls the device to enable the AI ​​system to identify user inputs, the processor load and power load associated with the AI ​​system can be reduced.

[0117] The method continues at step 1130, where the AI ​​system uses the motion sensor data samples to identify different user inputs to the system. The AI ​​system can input the motion sensor data samples into one or more AI models to identify or predict the user's desired input (e.g., by selecting the input with the highest probability). It will be appreciated that in some examples, the motion sensor data samples are filtered to extract features for use in identifying user inputs (e.g., gestures). Thus, in these examples, the AI ​​system is configured to receive the motion sensor data samples in the form of extracted features. The AI ​​system is configured to output the identified gestures for use by components of the system. For example, the identified user inputs can be used to control aspects of the system, such as an aerosol generator or a display. In this manner, the process shown in FIG. 11 provides an AI-supported user input function.

[0118] The AI-supported user input functionality described herein can be implemented in a wide range of devices, including flammable aerosol delivery systems, non-flammable aerosol delivery systems, or aerosol-free delivery systems.

[0119] As described herein, by embedding 3D gyroscope and accelerometer sensors (and / or other sensors for detecting movement) into electronic aerosol delivery systems or devices (e.g., on the circuit board of such devices), compact machine learning models can be trained and deployed to recognize consumer gestures (based on the movement data from the 3D gyroscope and accelerometer sensors) to complement or even completely replace mechanical movements. Advantageously, using filters to generate features, as described herein, can significantly reduce the amount of data processed by the AI ​​system, thereby speeding up processing and gesture identification by the AI ​​system.

[0120] Thus, an electronic aerosol delivery system is described that includes a motion sensor configured to provide data samples related to motion within the system, a filter configured to extract features from the motion sensor data samples, and an artificial intelligence (AI) system configured to receive the extracted features and use the extracted features to identify different user inputs to the system. The motion within the system can be motion of at least a portion of the system. For example, it can be motion of the entire system, where the electronic aerosol delivery device itself includes the filter and the AI ​​system, or it can be motion of an electronic aerosol delivery device in a system that includes the electronic aerosol delivery device and an external computing device that includes the filter and / or the AI ​​system.

[0121] Also described is an electronic aerosol delivery system that includes a motion sensor configured to provide data samples regarding motion within the system, an artificial intelligence (AI) system configured to receive the data samples and use the data samples to identify different user inputs for the system, and a user input function configured to control when the AI ​​system is used to identify the different user inputs. The motion within the system can be motion of at least a portion of the system. For example, it can be motion of the entire system, where the electronic aerosol delivery device itself includes a filter and an AI system, or it can be motion of an electronic aerosol delivery device in a system that includes the electronic aerosol delivery device and an external computing device that includes the filter and / or the AI ​​system.

[0122] The various embodiments described herein are presented solely to aid in the understanding and teaching of the claimed features. These embodiments are provided only as a representative sample of embodiments and are not intended to be exhaustive and / or exclusive. It is to be understood that the advantages, embodiments, examples, functions, features, structures, and / or other aspects described herein are not intended to be limitations on the scope of the invention as defined by the claims or equivalents thereof, and that other embodiments may be utilized and modifications may be made without departing from the scope of the claimed invention. Various embodiments of the present invention may suitably comprise, consist of, or consist essentially of any suitable combination of the disclosed elements, components, features, parts, steps, means, etc., other than those specifically described herein. In addition, the present disclosure may include other inventions not currently claimed but which may be claimed in the future.

Claims

1. a motion sensor configured to provide data samples relating to the motion of at least a portion of the electronic aerosol delivery system; a filter configured to extract features from the motion sensor data samples; an artificial intelligence (AI) system configured to receive the extracted features and use the extracted features to identify distinct user inputs to the electronic aerosol delivery system; An electronic aerosol delivery system comprising:

2. The electronic aerosol delivery system of claim 1 , wherein the data samples relating to the motion of the electronic aerosol delivery system provide a time series indicative of at least one of a position, a velocity, and / or an acceleration of the electronic aerosol delivery system.

3. 3. The electronic aerosol delivery system of claim 1 or 2, wherein the filter is provided as part of the output of the motion sensor, as part of the input of the AI ​​system, and / or as an intermediate component between the motion sensor and the AI ​​system.

4. The electronic aerosol delivery system of claim 1 , wherein the filter operates in the time domain and / or the frequency domain.

5. The electronic aerosol delivery system of claim 1 , wherein each extracted feature of the extracted features comprises a value of a statistical or mathematical property of the motion sensor data sample.

6. The electronic aerosol delivery system of claim 1 , wherein a buffer for storing the motion sensor data samples is maintained in a memory of the electronic aerosol delivery system.

7. 7. The electronic aerosol delivery system of claim 6, wherein the motion sensor is configured to acquire the motion sensor data samples at a first data rate, and the motion sensor is configured to provide the motion sensor data samples by incrementally storing the motion sensor data samples in the buffer at the first data rate.

8. The electronic aerosol delivery system of claim 7 , wherein the filter is configured to extract features at a second data rate, the second data rate being lower than the first data rate.

9. The electronic aerosol delivery system of any one of claims 1 to 8, wherein a first set of one or more features is extracted from the motion sensor data sample at a first state of the electronic aerosol delivery system, and a second, different set of one or more features is extracted from the motion sensor data sample at a second state of the electronic aerosol delivery system.

10. 10. The electronic aerosol delivery system of claim 9, wherein the first state corresponds to the electronic aerosol delivery system being in a quiescent state and the second state corresponds to the electronic aerosol delivery system being in an active state.

11. The electronic aerosol delivery system of any one of claims 1 to 10, wherein the electronic aerosol delivery system supports updating the features extracted from the data sample by reconfiguring the filter to correct and / or complement the features.

12. The electronic aerosol delivery system of claim 11 , wherein the electronic aerosol delivery system is configured to communicate with an external computing device to update the characteristics.

13. The electronic aerosol delivery system of claim 11 or 12, wherein the features can be updated to provide at least one of enhanced recognition, support for additional operating modes, and support for additional input symbols in a local language.

14. The electronic aerosol delivery system of any one of claims 1 to 13, wherein the AI ​​system comprises a plurality of AI models, each AI model responding to a different user input to the electronic aerosol delivery system.

15. The electronic aerosol delivery system of claim 14 , wherein each AI model is trained to identify a respective one of the different user inputs to the electronic aerosol delivery system.

16. The electronic aerosol delivery system of claim 15 , wherein the AI ​​system is configured to provide the extracted features to each of the AI ​​models to identify each of the different user inputs to the electronic aerosol delivery system.

17. The electronic aerosol delivery system of any one of claims 1 to 16, comprising an electronic aerosol delivery device comprising the motion sensor and, optionally, a controller configured to support the filter and / or the AI ​​system.

18. 18. The electronic aerosol delivery system of claim 17, comprising an external computing device configured to support the filter and / or the AI ​​system.

19. 19. The electronic aerosol delivery system of claim 17 or 18, wherein the electronic aerosol delivery device comprises a control unit for use in the electronic aerosol delivery system in conjunction with a cartridge containing an aerosol precursor.

20. 1. A method of operating an electronic aerosol delivery system, comprising: obtaining data samples relating to the movement of at least a portion of the electronic aerosol delivery system from a motion sensor; filtering the data samples to extract features of the data samples; using an artificial intelligence (AI) system to identify distinct user inputs to the electronic aerosol delivery system from the characteristics; A method comprising:

Images