Control method of intelligent lock induction handle, intelligent lock induction handle and intelligent lock
By deploying multiple sets of capacitive sensing PAD arrays and machine learning models on the smart lock handle, the problems of recognition accuracy, environmental adaptability and power consumption of the smart lock handle sensing solution are solved, achieving efficient door opening intention recognition and low power consumption control.
Patent Information
- Application Number
- CN202511900397.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-16
- Publication Date
- 2026-04-10
AI Technical Summary
Existing smart lock handle sensing solutions suffer from problems such as low accuracy in recognizing door opening intentions, poor environmental adaptability, insufficient adaptation to user habits, and unreasonable power consumption control.
An array of multiple capacitive sensing PADs is used, combined with a signal acquisition and preprocessing module, a microcontroller unit, and a communication module. A machine learning model is used to analyze the capacitance change data in real time to determine the user's intention to open the door, and power consumption is dynamically adjusted through a low-power wake-up mode and a high-frequency sampling mode.
It significantly improves the accuracy of door opening intention recognition, enhances environmental adaptability, achieves dynamic adaptation to user habits, and effectively reduces power consumption and extends battery life.
Smart Images
Figure CN121838306A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of intelligent lock, and in particular to a control method of an intelligent lock sensing handle, an intelligent lock sensing handle and an intelligent lock. BACKGROUND
[0002] With the rapid development of the smart home industry, as the core equipment of home security and convenient access, intelligent locks have gradually replaced traditional mechanical locks and become mainstream consumer products. The core requirement of intelligent locks is the balance between "safety" and "convenience", and the handle, as a key component for user interaction with the door lock, directly affects user experience and door lock security.
[0003] At present, the sensing trigger mode of the intelligent lock handle mainly includes the following two types: Mechanical trigger type: relying on the mechanical structure linkage (such as microswitch, connecting rod mechanism) when the user presses the handle or rotates the handle, triggering the door lock main control to wake up and execute the unlocking process. The core problem of this scheme is that the mechanical structure is easy to wear and has a limited service life, and it cannot distinguish between "accidental touch" and "real door opening intention", for example, when the user accidentally collides with the handle or children play and touch, it may trigger unnecessary wake-up or unlocking actions, which not only increases power consumption, but also has safety hazards.
[0004] Single sensor sensing type: using single type sensors such as infrared sensors, single point capacitive sensors or Hall sensors to detect "hand blocking infrared light", "single point capacitive change" or "magnetic field change" to determine whether the user is close to the handle. Although this scheme avoids mechanical wear, it has the following obvious limitations: Infrared sensors are easily affected by environmental light (such as strong light, backlight), temperature, and the misjudgment rate is significantly increased in summer high temperature or winter low temperature environment.
[0005] Single point capacitive sensors can only detect "whether there is an object close by", and cannot identify hand gestures (such as finger holding the handle, palm touching the handle), making it difficult to distinguish between "accidental touch" (such as clothes brushing) and "intentional door opening" (such as hand holding).
[0006] The Hall sensor needs to be used with a specific magnet, and the sensing distance is short, so the user needs to touch the handle to trigger it, which is not convenient.
[0007] In summary, the current intelligent lock handle sensing scheme mainly has the following technical problems: (1) Low accuracy of door opening intention recognition: unable to effectively distinguish between "accidental touch" (such as clothes brushing, collision) and "intentional door opening" (such as hand holding, finger pressing), resulting in false triggering (false wake-up, false unlocking) or missed triggering (user intentionally opens the door but does not trigger), affecting safety and convenience.
[0008] (2) Poor environmental adaptability: susceptible to environmental light, temperature, humidity and other factors, the sensing performance is greatly reduced in extreme environments such as strong light, high temperature and humidity, and the misjudgment rate is increased.
[0009] (3) Poor user habit adaptability: lack of self-learning ability, unable to dynamically adjust the sensing logic according to different user hand gestures and holding habits, and the experience is inconsistent when used by different users.
[0010] (4) Unreasonable power consumption control: in order to ensure the sensing coverage, the sensor needs to be kept in high frequency wake-up, which leads to high overall power consumption of the smart lock, short battery life, and increased frequency and cost of replacing the battery for the user.
[0011] Therefore, the prior art still needs to be improved and developed. SUMMARY
[0012] The main purpose of the present application is to provide a control method for a smart lock sensing handle, a smart lock sensing handle and a smart lock, aiming to solve the problems of low door opening intention recognition accuracy, poor environmental adaptability, insufficient user habit adaptability and unreasonable power consumption control in the prior art.
[0013] To achieve the above purpose, the present application provides a control method for a smart lock sensing handle, the smart lock sensing handle comprising: a capacitive sensing array module, a signal acquisition and preprocessing module, a micro control unit and a communication module, the capacitive sensing array module comprising a plurality of groups of capacitive sensing PADs, the control method for the smart lock sensing handle comprising the following steps: When the smart lock is powered on, the capacitive sensing array module is in a low-power wake-up mode by default, and only when a certain group of capacitive sensing PADs detects that the capacitive change exceeds a preset threshold, it is switched to a high-frequency sampling mode for data acquisition; The signal acquisition and preprocessing module filters the multiple groups of PAD data sampled at high frequency, and transmits the filtered multiple groups of PAD capacitive change data to the micro control unit; The micro control unit runs a trained machine learning model, and analyzes the multiple groups of PAD capacitive change data in real time through the machine learning model to determine whether there is a door opening intention; If there is a door opening intention, the micro control unit sends an unlock trigger signal to the smart lock master control through the communication module, and the smart lock master control controls unlocking according to the unlock trigger signal.
[0014] Optionally, the control method of the intelligent lock sensing handle, wherein the capacitive sensing array module is in a low-power wake-up mode by default after the intelligent lock is powered on, and is switched to a high-frequency sampling mode for data collection only when a certain group of capacitive sensing PADs detects a capacitive change amount exceeding a preset threshold, and further comprising: The number of capacitive sensing PADs is set in advance according to the area size of the handle top, the handle inner side, and the handle side of the intelligent lock sensing handle, and the distance between adjacent capacitive sensing PADs is a preset distance.
[0015] Optionally, the control method of the intelligent lock sensing handle, wherein the training process of the machine learning model comprises: Collecting positive samples and negative samples in collection scenes covering different environments and different user samples, wherein the positive samples are capacitive change sequence data of users with door opening intention, and the negative samples are capacitive change sequence data of users without door opening intention; Labeling the positive samples as intention labels and labeling the negative samples as no-intention labels to form a training data set; According to each group of capacitive change sequence data in the training data set, extracting time domain features, spatial features, and posture features; Using a light-weight support vector machine or a random forest model as a machine learning model, inputting the time domain features, the spatial features, and the posture features into the machine learning model for training, optimizing the model parameters, and quantifying the trained machine learning model to obtain the trained machine learning model.
[0016] Optionally, the control method of the intelligent lock sensing handle, wherein the time domain features include the maximum value, the minimum value, the mean value, the variance, and the change rate of each group of PAD capacitive changes, the spatial features include the capacitive change correlation between different PADs, and the posture features include the duration and the change trend of the capacitive change sequence.
[0017] Optionally, the control method of the intelligent lock sensing handle, wherein the filtering process includes mean filtering and Kalman filtering to remove environmental electromagnetic interference and transient noise.
[0018] Optionally, the control method of the intelligent lock sensing handle, wherein the control method of the intelligent lock sensing handle further comprises: The micro-control unit records the capacitive data of the user each time the door is successfully opened, and fine-tunes the machine learning model every preset period.
[0019] Optionally, the control method of the intelligent lock sensing handle, wherein the judgment of whether there is a door opening intention further comprises: If there is no door opening intention, the micro control unit controls the capacitive sensing array module to switch from the high frequency sampling mode to the low power consumption wake-up mode, and does not send any signal.
[0020] In addition, to achieve the above-mentioned purpose, the application further provides a smart lock induction handle, which comprises a capacitive sensing array module, a signal acquisition and preprocessing module, a micro control unit and a communication module, the capacitive sensing array module comprises a plurality of groups of capacitive sensing PADs; When the smart lock is powered on, the capacitive sensing array module is used to be in the low power consumption wake-up mode by default, and only when a certain group of capacitive sensing PADs detects that the capacitive change exceeds the preset threshold, it is switched to the high frequency sampling mode for data acquisition; The signal acquisition and preprocessing module is used to filter the high frequency sampling of the plurality of groups of PAD data, and transmit the filtered plurality of groups of PAD capacitive change data to the micro control unit; The micro control unit is used to run the trained machine learning model, and analyzes the plurality of groups of PAD capacitive change data in real time through the machine learning model to determine whether there is a door opening intention; If there is a door opening intention, the micro control unit is also used to send an unlock trigger signal to the smart lock master through the communication module.
[0021] Optionally, the smart lock induction handle, wherein the capacitive sensing PAD is a capacitive sensing PAD made of FPC flexible material, which is attached to the top, inner side and side of the handle of the smart lock induction handle.
[0022] In addition, to achieve the above-mentioned purpose, the application further provides a smart lock, which comprises the smart lock induction handle, and further comprises a smart lock master, which is used to control unlocking according to the unlock trigger signal.
[0023] In this invention, the smart lock sensor handle includes: a capacitive sensing array module, a signal acquisition and preprocessing module, a microcontroller unit, and a communication module. The capacitive sensing array module includes multiple sets of capacitive sensing PADs. The method includes: when the smart lock is powered on, the capacitive sensing array module is in a low-power wake-up mode by default. Only when a set of capacitive sensing PADs detects a capacitance change exceeding a preset threshold, it switches to a high-frequency sampling mode for data acquisition. The signal acquisition and preprocessing module filters the high-frequency sampled data from multiple sets of PADs and transmits the filtered data on the capacitance changes of the multiple sets of PADs to the microcontroller unit. The microcontroller unit runs a trained machine learning model to analyze the capacitance change data of the multiple sets of PADs in real time to determine whether there is an intention to open the door. If there is an intention to open the door, the microcontroller unit sends an unlock trigger signal to the smart lock master controller through the communication module, and the smart lock master controller controls the unlocking according to the unlock trigger signal. This invention deploys a sensor array consisting of multiple capacitive sensing PADs on the smart lock's sensor handle to collect data on the capacitance changes when a user's hand approaches or places their hand on the door handle. The collected capacitance change data is then trained and analyzed in real time using a machine learning model to accurately determine whether the user intends to open the door, significantly improving the accuracy of door opening intention recognition. Attached Figure Description
[0024] Figure 1 This is a structural diagram of a preferred embodiment of the control system for the intelligent lock sensor handle of the present invention; Figure 2 This is a schematic diagram of the layout design of the capacitive sensing array in a preferred embodiment of the control system of the intelligent lock sensor handle of the present invention. Figure 3 This is a flowchart of a preferred embodiment of the control method for the intelligent lock sensor handle of the present invention; Figure 4 This is a flowchart illustrating the process of determining whether there is an intention to open the door in a preferred embodiment of the control method for the intelligent lock sensor handle of the present invention. Detailed Implementation
[0025] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
[0026] The smart lock sensor handle described in the preferred embodiment of the present invention, such as... Figure 1As shown, the smart lock sensor handle includes: a capacitive sensing array module, a signal acquisition and preprocessing module, a microcontroller unit, and a communication module. The capacitive sensing array module includes multiple sets of capacitive sensing PADs (Point of Action / Activation Device, which are usually referred to as "sensing pads" or "electrodes" in the field of touch technology. Essentially, it is a capacitive electrode used to form a coupling capacitance with the user's finger (or stylus) to detect touch or proximity actions).
[0027] The capacitive sensing PAD is a flexible material made of FPC (Flexible Printed Circuit, a type of printed circuit board made of flexible insulating substrate (such as polyimide PI) with high reliability and excellent flexibility), and is attached to the top, inner side, and sides of the smart lock's sensor handle (i.e., the areas that the user's hand is likely to touch when holding the handle). Figure 2 As shown, the specific layout principles are as follows: Deploy 6-12 sets of sensor PADs depending on the handle length (usually 5-15cm) to ensure that a single PAD can detect local contact of a finger or palm.
[0028] Coverage area: Top of handle (1-3 PADs): Detects user finger touches on the top (some user habits); Inside of the handle (3-6 PADs): Covers the main area of hand grip, detecting the contact area between the palm and the handle; Handle side (2-3 PADs): Detects finger wrapping action on the handle (such as thumb pressing the side, other fingers gripping the inside side); Spacing: The spacing between adjacent PADs is 3-5mm (i.e., the preset distance between adjacent capacitive sensing PADs is 3-5mm) to avoid signal crosstalk and ensure no sensing dead zones.
[0029] Specifically, when the smart lock is powered on, the capacitive sensing array module is in a low-power wake-up mode by default. Only when a group of capacitive sensing PADs detects a capacitance change exceeding a preset threshold, it switches to a high-frequency sampling mode for data acquisition. The signal acquisition and preprocessing module (including a conditioning chip and filtering circuit for acquiring the capacitance signal of the capacitive sensing array module) is used to filter the high-frequency sampled PAD data and transmit the filtered PAD capacitance change data to the microcontroller unit (low-power MCU). The microcontroller unit is used to run a trained machine learning model to analyze the PAD capacitance change data in real time and determine whether there is an intention to open the door. If there is an intention to open the door, the microcontroller unit is also used to send an unlock trigger signal to the smart lock master controller through the communication module (I2C / SPI / UART communication interface).
[0030] The smart lock described in the preferred embodiment of the present invention, in addition to including the aforementioned smart lock sensor handle, includes, for example... Figure 1 As shown, the smart lock also includes a smart lock master controller, which is used to control unlocking based on the unlock trigger signal.
[0031] Furthermore, the control method for the smart lock sensor handle described in the preferred embodiment of the present invention, such as... Figure 3 and Figure 4 As shown, the control method for the smart lock's sensor handle includes the following steps: Step S10: When the smart lock is powered on, the capacitive sensing array module is in low-power wake-up mode by default. Only when a certain group of capacitive sensing PADs detects a change in capacitance exceeding a preset threshold will it switch to high-frequency sampling mode for data acquisition.
[0032] Specifically, the number of capacitive sensing PADs is pre-set according to the area of the top, inner, and side surfaces of the smart lock's sensor handle, with a preset distance between adjacent capacitive sensing PADs. When the smart lock is powered on, the capacitive sensing array module is in a low-power wake-up mode by default (PAD sampling frequency of 5~10Hz). Only when a group of PADs detects a capacitance change exceeding a preset threshold (such as 5% of the baseline) will it switch to a high-frequency sampling mode (sampling frequency increased to 100Hz).
[0033] Here, "baseline" refers to the reference value. The steady-state value of a capacitive sensor installed in a defined environment, without interference from external conductive materials, is the baseline. For example, a capacitive PAD mounted on a handle might have a reference value of 22. Without external interference, it will remain around 22 for a long time. When a hand approaches, the value increases from 22, reaching 56 when the hand is fully against the handle. By analyzing the changes in the capacitive sensing value, we can determine the trend of an increasing and then sustained high value when a conductive object approaches the PAD. Generally, a 5% to 20% increase is added to the reference value as the trigger threshold, depending on the fluctuation range of the sensing value.
[0034] Step S20: The signal acquisition and preprocessing module filters the multiple sets of PAD data sampled at high frequency and transmits the filtered multiple sets of PAD capacitance change data to the microcontroller unit.
[0035] Specifically, the signal acquisition and preprocessing module filters multiple sets of PAD data sampled at high frequency (for example, using mean filtering + Kalman filtering to remove environmental electromagnetic interference and instantaneous noise; mean filtering is a very intuitive time-domain static filtering method, whose basic assumption is that noise is random while the signal is relatively stable; by taking the arithmetic mean of multiple measurements within a time window, it is expected that positive and negative noise will cancel each other out, thereby approximating the real signal; Kalman filtering is an optimal recursive estimation algorithm based on a state-space model; its core idea is to combine the two steps of "prediction" and "update" and use the dynamic model of the system and measurement data to make an optimal estimate of the true state of the system), and transmits the filtered multiple sets of PAD capacitance change data to the microcontroller unit (MCU).
[0036] Step S30: The microcontroller unit runs the trained machine learning model and analyzes the multiple sets of PAD capacitance change data in real time to determine whether there is an intention to open the door.
[0037] Specifically, before using a pre-trained machine learning model for analysis, the machine learning model needs to be trained first. The training process of the machine learning model includes: (1) Data collection: Positive and negative samples were collected in different environments (normal temperature, high temperature 45℃, low temperature -10℃, humid) and different user samples (20-50 years old, including left-handed users, right-handed users, and different hand sizes).
[0038] The positive samples are capacitance change sequence data of users with the intention to open the door. Positive samples (with the intention to open the door) include various typical door opening postures such as normal gripping of the handle, pressing the handle with fingers, and palm touching the handle. For each posture, 500 sets of multi-PAD capacitance change sequence data are collected (each set of sequences contains the capacitance values of 6-12 PADs, and the sampling time is 1 second).
[0039] The negative samples are capacitance change sequence data when the user has no intention to open the door. The negative samples (no intention to open the door) include various typical scenarios such as clothing wiping the handle, objects hitting the handle, fingers accidentally touching the handle, and environmental noise (such as electromagnetic interference). For each scenario, 500 sets of PAD capacitance change data are collected.
[0040] Data labeling: The positive samples are labeled with intentional labels, and the negative samples are labeled with unintentional labels to form a training dataset.
[0041] (2) Feature extraction: Based on each set of capacitance change sequence data in the training dataset, extract three key features (ensuring that the features can distinguish between "intentional" and "unintentional" actions). The three features are: Time-domain characteristics: maximum, minimum, mean, variance, and rate of change of capacitance for each group of PADs (e.g., capacitance change within 50ms). Spatial characteristics: the correlation of capacitance changes between different PADs (such as the synchronicity of changes between the PADs on the inside of the handle and the top PAD; when intentionally gripping the handle, the hand grip will cause multiple PADs to change synchronously; when unintentionally gripping the handle, only a single PAD may show a momentary change). Postural characteristics: duration of capacitance change sequence (when intentional, hand gripping action lasts 0.3-1s; when unintentional, action lasts <0.3s), and trend of change (when intentional, capacitance value shows a "rapid increase-stable" trend; when unintentional, it shows a "momentary increase-rapid decrease" trend).
[0042] (3) Model training and optimization: Model selection: Use lightweight "Support Vector Machine (SVM)" or "Random Forest" models (adapted to the computing power of low-power MCUs and avoiding the latency caused by complex models) as the machine learning model.
[0043] Training process: The extracted three types of features are input as the machine learning model. Five-fold cross-validation (a common model evaluation and parameter optimization method in machine learning, whose core purpose is to more reliably test the model's "generalization ability" with a limited dataset, verifying the model's performance on different datasets and in different scenarios; usually, the data is split, for example, into 5 parts, to split the training set and the validation set, using the training set to train the model and the validation set to validate and evaluate the model's performance; by continuously splitting the training set and the validation set, and through multiple rounds of training and validation, the average of multiple validations is taken as the final evaluation result of the model) is used to optimize the model parameters (such as the kernel function parameters of SVM, the number of decision trees in random forest) to ensure the model's generalization ability on different environments and user data.
[0044] Model compression: The trained machine learning model is quantized (e.g., 32-bit floating-point quantization is converted into 16-bit fixed-point quantization) to reduce the model size (the compressed model size is <100KB) and ensure that it can run efficiently in the MCU, thereby obtaining the trained machine learning model.
[0045] The extracted real-time features are input into a pre-trained machine learning model, which determines whether there is an intention to open the door and outputs a judgment result of "there is an intention to open the door" or "there is no intention to open the door". The judgment time is <300ms.
[0046] Step S40: If there is an intention to open the door, the microcontroller sends an unlock trigger signal to the smart lock master controller through the communication module, and the smart lock master controller controls the unlocking according to the unlock trigger signal.
[0047] Specifically, if the judgment result is "intent to open the door": the microcontroller unit (MCU) sends an unlock trigger signal to the smart lock master controller through the communication module. The unlock trigger signal includes "intent confidence level". If the confidence level is >90%, unlocking is triggered directly.
[0048] If there is no intention to open the door, i.e., the judgment result is "no intention to open the door", the microcontroller unit (MCU) controls the capacitive sensing array module to switch from the high-frequency sampling mode to the low-power wake-up mode, and does not send any signal.
[0049] Furthermore, the smart lock supports user-defined learning. The microcontroller unit can record the capacitance data each time the user "successfully opens the door" and fine-tune the machine learning model periodically (e.g., every 7 days) to optimize its adaptability to the user's usage habits. The update process is completed locally on the MCU.
[0050] This invention deploys a sensor array consisting of multiple capacitive sensing PADs on the smart lock's sensor handle to collect data on the capacitance changes when a user's hand approaches or places their hand on the door handle. The collected capacitance change data is trained and analyzed in real time using a machine learning model to accurately determine whether the user intends to open the door. When an intention to open the door is determined, the smart lock's sensor handle sends an unlock signal to the smart lock's main controller, triggering the subsequent unlocking process.
[0051] The technical effects that this invention can bring are as follows: (1) Significantly improve the accuracy of door opening intention recognition: The model judges the unlocking intention through multi-dimensional features, avoiding the risk of misunderstanding the lock and reducing missed triggers.
[0052] (2) Enhanced environmental adaptability: Capacitive sensing technology is not affected by ambient light and temperature (the change in capacitance is only related to the dielectric constant of the object, and the dielectric constant of the hand is much higher than that of air and clothing. The effect of ambient temperature on capacitance can be offset by the filter circuit).
[0053] (3) Achieve dynamic adaptation to user habits: The machine learning model supports adaptive updates and can learn the grip habits of different users (such as the PAD change sequence of left-handed users and the action characteristics of the fingers touching the top first) and adjust the sensing sensitivity. For example, for users who are used to gripping with their left hand, the weight of the PAD on the left side of the handle can be increased to ensure that users can trigger the device without changing their habits; reduce the consistency error of trigger position and trigger force when different users use the device, and significantly improve the user experience.
[0054] (4) Effectively reduce power consumption and extend battery life: The dynamic power consumption control strategy of low power wake-up mode + high frequency trigger sampling is adopted. In low power mode, the PAD sampling frequency is only 5~10Hz, the MCU is in sleep state, and the power consumption is <40μA; it only switches to high frequency mode (power consumption <500μA) when a potential contact is detected, and the high frequency mode lasts for only 1s.
[0055] In summary, this invention provides a control method for a smart lock sensor handle, a smart lock sensor handle, and a smart lock. The smart lock sensor handle includes a capacitive sensing array module, a signal acquisition and preprocessing module, a microcontroller unit, and a communication module. The capacitive sensing array module includes multiple sets of capacitive sensing PADs. The method includes: when the smart lock is powered on, the capacitive sensing array module is in a low-power wake-up mode by default. Only when a set of capacitive sensing PADs detects a capacitance change exceeding a preset threshold, it switches to a high-frequency sampling mode for data acquisition. The signal acquisition and preprocessing module filters the high-frequency sampled data from the multiple sets of PADs and transmits the filtered data on the capacitance changes of the multiple sets of PADs to the microcontroller unit. The microcontroller unit runs a trained machine learning model to analyze the capacitance change data of the multiple sets of PADs in real time to determine whether there is an intention to open the door. If there is an intention to open the door, the microcontroller unit sends an unlock trigger signal to the smart lock master controller through the communication module, and the smart lock master controller controls the unlocking according to the unlock trigger signal. This invention deploys a sensor array consisting of multiple capacitive sensing PADs on the smart lock's sensor handle to collect data on the capacitance changes when a user's hand approaches or places their hand on the door handle. The collected capacitance change data is then trained and analyzed in real time using a machine learning model to accurately determine whether the user intends to open the door, significantly improving the accuracy of door opening intention recognition.
[0056] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal that includes that element.
[0057] Of course, those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.). The program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The computer-readable storage medium can be a memory, magnetic disk, optical disk, etc.
[0058] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.
Claims
1. A control method for a smart lock sensor handle, characterized in that, The smart lock sensor handle includes: a capacitive sensing array module, a signal acquisition and preprocessing module, a microcontroller unit, and a communication module. The capacitive sensing array module includes multiple sets of capacitive sensing PADs. The control method of the smart lock sensor handle includes: When the smart lock is powered on, the capacitive sensing array module is in low-power wake-up mode by default. It only switches to high-frequency sampling mode to collect data when a certain group of capacitive sensing PADs detects a change in capacitance exceeding a preset threshold. The signal acquisition and preprocessing module filters multiple sets of PAD data sampled at high frequency and transmits the filtered multiple sets of PAD capacitance change data to the microcontroller unit. The microcontroller unit runs a trained machine learning model, which analyzes the multiple sets of PAD capacitance change data in real time to determine whether there is an intention to open the door. If there is an intention to open the door, the microcontroller sends an unlock trigger signal to the smart lock master controller through the communication module, and the smart lock master controller controls the unlocking according to the unlock trigger signal.
2. The control method for the intelligent lock sensor handle according to claim 1, characterized in that, When the smart lock is powered on, the capacitive sensing array module is in low-power wake-up mode by default. It only switches to high-frequency sampling mode for data acquisition when a group of capacitive sensing PADs detects a capacitance change exceeding a preset threshold. This also includes the following: The number of capacitive sensing PADs is pre-set according to the area size of the top, inner and side surfaces of the smart lock's sensor handle, and the spacing between adjacent capacitive sensing PADs is a preset distance.
3. The control method for the intelligent lock sensor handle according to claim 1, characterized in that, The training process of the machine learning model includes: Positive and negative samples are collected in collection scenarios covering different environments and different user samples. The positive samples are capacitance change sequence data when the user intends to open the door, and the negative samples are capacitance change sequence data when the user does not intend to open the door. The positive samples are labeled with intentional labels, and the negative samples are labeled with unintentional labels to form a training dataset; Based on each set of capacitance change sequence data in the training dataset, extract temporal features, spatial features, and attitude features; A lightweight support vector machine or random forest model is used as the machine learning model. The temporal features, spatial features, and pose features are input into the machine learning model for training. The model parameters are optimized, and the trained machine learning model is quantized to obtain the trained machine learning model.
4. The control method for the intelligent lock sensor handle according to claim 3, characterized in that, The temporal features include the maximum, minimum, mean, variance, and rate of change of capacitance for each group of PADs; the spatial features include the correlation of capacitance changes among different PADs; and the attitude features include the duration and trend of the capacitance change sequence.
5. The control method for the intelligent lock sensor handle according to claim 1, characterized in that, The filtering process includes mean filtering and Kalman filtering to remove environmental electromagnetic interference and transient noise.
6. The control method for the intelligent lock sensor handle according to claim 1, characterized in that, The control method for the smart lock's sensor handle also includes: The microcontroller records the capacitance data each time the user successfully opens the door, and fine-tunes the machine learning model every preset period.
7. The control method for the intelligent lock sensor handle according to claim 1, characterized in that, The determination of whether there is an intention to open the door also includes: If there is no intention to open the door, the microcontroller controls the capacitive sensing array module to switch from the high-frequency sampling mode to the low-power wake-up mode without sending any signal.
8. A smart lock sensor handle, characterized in that, The smart lock's sensor handle includes: a capacitive sensing array module, a signal acquisition and preprocessing module, a microcontroller unit, and a communication module. The capacitive sensing array module includes multiple sets of capacitive sensing PADs. When the smart lock is powered on, the capacitive sensing array module is in low-power wake-up mode by default. It only switches to high-frequency sampling mode to collect data when a group of capacitive sensing PADs detects a change in capacitance exceeding a preset threshold. The signal acquisition and preprocessing module is used to filter multiple sets of PAD data sampled at high frequency, and transmit the filtered multiple sets of PAD capacitance change data to the microcontroller unit. The microcontroller unit is used to run a trained machine learning model, which analyzes the multiple sets of PAD capacitance change data in real time to determine whether there is an intention to open the door. If there is an intention to open the door, the microcontroller unit is also used to send an unlocking trigger signal to the smart lock master controller through the communication module.
9. The smart lock sensor handle according to claim 8, characterized in that, The capacitive sensing PAD is made of FPC flexible material and is attached to the top, inner side and side of the smart lock's sensor handle.
10. A smart lock, characterized in that, The smart lock includes the smart lock sensor handle as described in claim 8, and the smart lock also includes a smart lock master controller, which is used to control unlocking according to an unlock trigger signal.