Gesture recognition control method and device for air conditioner, electronic equipment and air conditioner

Abstract

The embodiment of the invention provides a gesture recognition control method and device for an air conditioner, electronic equipment and the air conditioner. The method comprises the steps of obtaining amplitude data of multiple pieces of first channel state information within a first preset duration, and determining a first dispersion statistical feature of the amplitude data of the multiple pieces of first channel state information, the first dispersion statistical feature comprising at least one of a median absolute deviation and a quartile distance; in response to the situation that the current first dispersion statistical feature is greater than or equal to a first wake-up threshold, amplitude data of second channel state information within a second preset duration is obtained, and the second preset duration comprises the first preset duration; and whether the gesture recognition function of the air conditioner is awakened or not is determined according to the amplitude data of the second channel state information. According to the air conditioner control system, power consumption of the air conditioner control system is reduced, and low-power-consumption standby is achieved.

Description

Technical Field

[0001] This disclosure relates to the field of air conditioning technology, and in particular to a gesture recognition control method and device for an air conditioner, an electronic device, and an air conditioner. Background Technology

[0002] In current technology, air conditioners are controlled by remote controls, which can be difficult to use, especially in dark environments like nighttime. Some air conditioners have voice control functions, and a very small number have camera functions, but due to increased costs and serious privacy issues, these air conditioners are often not popular with consumers. Summary of the Invention

[0003] This disclosure provides a gesture recognition control method and apparatus for an air conditioner, an electronic device, and an air conditioner, to solve or alleviate one or more technical problems in the prior art.

[0004] As a first aspect of the present disclosure, the present disclosure provides a gesture recognition control method for an air conditioner, including: S11: Every first preset time interval, acquire amplitude data of multiple first channel state information within a first preset duration, and determine the first discrete statistical feature of the amplitude data of multiple first channel state information. The first discrete statistical feature includes at least one of median absolute deviation and interquartile range. S12: In response to the current first discrete statistical feature being greater than or equal to the first wake-up threshold, obtain the amplitude data of the second channel state information within the second preset duration. The second preset duration is the duration between the first moment and the second moment. The first moment is before the start time of the current first preset duration, and the second moment is after the start time of the current second preset duration. The first wake-up threshold represents the discrete statistical feature of the amplitude data of the channel state information in the resting state. S13: Determine whether to wake up the gesture recognition function of the air conditioner based on the amplitude data of the second channel state information.

[0005] In some embodiments, the first preset duration includes N consecutive time slices, the first discrete statistical feature includes median absolute deviation, and the acquisition of amplitude data of multiple first channel state information within the first preset duration, and the determination of the first discrete statistical feature of the amplitude data of the multiple first channel state information, includes: For N consecutive time slices, obtain the amplitude data of p first channel state information in each time slice to construct an amplitude data group, which includes N*p amplitude data. Determine the median of the amplitude data set; The absolute deviation of the median is determined based on the median and the amplitude data of each first channel state information.

[0006] In some embodiments, the method further includes: a first wake-up threshold is calculated based on the mean and standard deviation of a discrete statistical feature sequence, the discrete statistical feature sequence including a plurality of second discrete statistical features, each of which is calculated from a plurality of amplitude data in the channel state information in the resting state.

[0007] In some embodiments, determining whether to activate the gesture recognition function of the air conditioner based on the amplitude data of the second channel state information includes: Based on the amplitude data of the second channel state information, determine the spectrum information of the second channel state information; Based on the spectrum information, determine whether to activate the air conditioner's gesture recognition function.

[0008] In some embodiments, determining the spectrum information of the second channel state information based on the amplitude data of the second channel state information includes: Short-time Fourier transform is performed on the amplitude data of the second channel state information to obtain the spectrum information.

[0009] In some embodiments, determining whether to activate the air conditioner's gesture recognition function based on spectrum information includes: Based on the spectrum information, the total spectrum energy within the preset trigger frequency range is determined. The preset trigger frequency range is used to characterize the action frequency range to which the trigger gesture of the air conditioner belongs. Based on the total spectrum energy, determine whether to activate the air conditioner's gesture recognition function.

[0010] In some embodiments, determining whether to activate the air conditioner's gesture recognition function based on the total spectral energy includes: Obtain the matching confidence level between the total spectral energy and the trigger gesture of the air conditioner; The gesture recognition function of the air conditioner is activated when the matching confidence level is greater than the second wake-up threshold.

[0011] In some embodiments, it also includes: In response to a matching confidence level less than or equal to a second wake-up threshold, the process returns to step S11 to obtain amplitude data of multiple first channel state information within the next first preset time period, and determines the first discrete statistical feature of the amplitude data of the multiple first channel state information.

[0012] In some embodiments, obtaining the matching confidence level between the total spectral energy and the trigger gesture of the air conditioner includes: The total spectral energy is input into a pre-trained trigger gesture binary classification model to obtain the matching confidence. The trigger gesture binary classification model is trained using multiple gesture samples and the confidence samples corresponding to each gesture sample.

[0013] In some embodiments, it also includes: Gesture feature recognition is performed on the second channel state information within the second preset time period to obtain gesture feature information; Determine control commands based on gesture characteristics; Adjust the operating status of the air conditioner according to the control commands.

[0014] In some embodiments, the method further includes: in response to the current first discrete statistical feature being less than the first wake-up threshold, returning to step S11, obtaining amplitude data of a plurality of first channel state information within the next first preset time period, and determining the first discrete statistical feature of the amplitude data of the plurality of first channel state information.

[0015] As a second aspect of this disclosure, this disclosure provides a gesture recognition control device for an air conditioner, including: A first processor is configured to acquire amplitude data of multiple first channel state information within a first preset time period, and determine a first discrete statistical feature of the amplitude data of the multiple first channel state information, wherein the first discrete statistical feature includes at least one of median absolute deviation and interquartile range. The second processor is used to acquire amplitude data of the second channel state information within a second preset duration when the current first discrete statistical feature is greater than the first wake-up threshold. The second preset duration is the duration between the first moment and the second moment, where the first moment is before the start time of the current first preset duration and the second moment is after the start time of the current second preset duration. The second processor is also used to determine whether to wake up the gesture recognition function of the air conditioner based on the amplitude data of the second channel state information.

[0016] In some embodiments, The second processor is also used to perform gesture feature recognition on the second channel state information within a second preset time period to obtain gesture feature information; and to determine control commands based on the gesture feature information. The device also includes an instruction execution module, which is used to adjust the operating status of the air conditioner according to control instructions.

[0017] As a third aspect of this disclosure, this disclosure provides an electronic device, characterized in that it includes: At least one processor; and A memory that is communicatively connected to at least one processor; wherein, The memory stores instructions that can be executed by at least one processor to enable the at least one processor to perform any of the methods disclosed herein.

[0018] As a fourth aspect of the present disclosure, the present disclosure provides an air conditioner that includes the control device of the present disclosure, or an electronic device that includes the present disclosure.

[0019] The technical solution of this disclosure embodiment, in standby mode, processes the amplitude data of multiple first channel state information within a first preset time period, and only when the current first discrete statistical feature is greater than or equal to the first wake-up threshold, does it need to process the amplitude data of second channel state information within a second preset time period. This method does not require continuous high-intensity extraction of CSI data stream, significantly reduces the power consumption of the air conditioner control system, and realizes a low-power standby mode.

[0020] The technical solution disclosed herein sets a first wake-up threshold. Only when the current first discrete statistical characteristic is greater than the first wake-up threshold is it necessary to acquire the amplitude data of the second channel state information within a second preset time period to determine whether to wake up the gesture recognition function. In other words, before waking up the gesture recognition function, a two-level judgment rule is set: not only must the current first discrete statistical characteristic be greater than or equal to the first wake-up threshold, but the amplitude data of the second channel state information must also be used to determine whether to wake up the gesture recognition function. This approach can reduce false wake-up caused by unconscious user actions, improve trigger recognition accuracy, reduce the false trigger rate, effectively eliminate transient environmental interference, and improve the robustness of the control system.

[0021] The above overview is for illustrative purposes only and is not intended to be limiting in any way. Further aspects, embodiments, and features of this disclosure will become readily apparent from the accompanying drawings and the following detailed description, in addition to the illustrative aspects, embodiments, and features described above. Attached Figure Description

[0022] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the various drawings denote the same or similar parts or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings depict only some embodiments according to this disclosure and should not be construed as limiting the scope of this disclosure.

[0023] Figure 1 This is a flowchart illustrating a gesture recognition control method for an air conditioner according to an embodiment of the present disclosure; Figure 2 This is a schematic diagram of the working process of the gesture recognition control device disclosed herein. Detailed Implementation

[0024] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the spirit or scope of this disclosure, and different embodiments can be combined arbitrarily without conflict. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0025] To facilitate the control of air conditioners, some technologies use cameras to capture human eye gaze signals and hand gestures for control. However, this method has some drawbacks. For example, prolonged camera operation may raise concerns about user privacy, cameras have difficulty recognizing signals in dark environments at night, and adding sensors increases costs.

[0026] Some related technologies exist for air conditioners that collect raw Channel State Information (CSI) to recognize gesture features and adjust the air conditioner's operation accordingly. However, this approach involves continuous, highly complex feature extraction (such as complete time-domain, frequency-domain, or subcarrier fusion processing) and complex model matching of the CSI data stream, causing the Wi-Fi module's main processor to run for extended periods, resulting in excessively high overall energy consumption of the air conditioning control system. Furthermore, it suffers from high rates of false triggering and misrecognition, potentially misinterpreting unintentional user actions (such as getting up, turning around, or pet activity) as control gestures, leading to frequent system malfunctions and severely impacting user experience.

[0027] Channel State Information (CSI) is an important concept in wireless communication, referring to the characteristic information of a wireless channel, including signal-to-noise ratio, multipath effects, and fading. CSI can be used for indoor positioning, activity recognition, gesture recognition, and more.

[0028] To facilitate easier control of air conditioners and reduce energy consumption, this disclosure provides a gesture recognition control method for air conditioners. In this method, amplitude data of multiple first channel state information within a first preset time interval are acquired every first preset time interval, and a first discrete statistical feature of the amplitude data of the multiple first channel state information is determined. Then, the first discrete statistical feature is compared with a first wake-up threshold. When the current first discrete statistical feature is greater than the first wake-up threshold, amplitude data of second channel state information within a second preset time interval is acquired. Finally, based on the amplitude data of the second channel state information, it is determined whether to wake up the gesture recognition function of the air conditioner. Before waking up the gesture recognition function of the air conditioner, the air conditioner performs CSI amplitude data acquisition and related calculations, which significantly reduces the standby power consumption of the air conditioner compared to CSI gesture recognition in related technologies. The technical solution of this disclosure is described in detail below through specific embodiments.

[0029] Figure 1 This is a flowchart illustrating a gesture recognition control method for an air conditioner according to an embodiment of this disclosure. Figure 1 As shown, the gesture recognition control method for air conditioners includes steps S11 to S13.

[0030] In step S11, the amplitude data of multiple first channel state information within a first preset time period is obtained, and the first discrete statistical feature of the amplitude data of multiple first channel state information is determined. The first discrete statistical feature includes at least one of median absolute deviation and interquartile range.

[0031] Air conditioners are equipped with CSI acquisition modules, such as Wi-Fi transceiver chips and their antennas. These CSI acquisition modules can collect CSI data from the indoor environment, thereby obtaining CSI amplitude data.

[0032] The first preset time interval is pre-set, and its specific value can be set as needed. The CSI acquisition module acquires amplitude data of multiple first channel state information within the first preset time period, thereby obtaining amplitude data A1, A2, ..., Am of multiple first channel state information. A1, A2, ..., Am can be denoted as an amplitude data group A, and this amplitude data group is denoted as A={A1,A2,...,Am}.

[0033] For example, a first discrete statistical feature of the amplitude data set can be determined using mathematical statistical methods. The first discrete statistical feature includes at least one of the median absolute deviation and interquartile range.

[0034] In step S12, in response to the current first discrete statistical feature being greater than or equal to the first wake-up threshold, the amplitude data of the second channel state information within the second preset duration is obtained. The second preset duration is the duration between the first moment and the second moment. The first moment is located before the start time of the current first preset duration, and the second moment is located after the start time of the current second preset duration. The first wake-up threshold represents the discrete statistical feature of the amplitude data of the channel state information in the resting state.

[0035] In step S13, based on the amplitude data of the second channel state information, it is determined whether to wake up the gesture recognition function of the air conditioner.

[0036] The resting state refers to the stable state of the Wi-Fi signal in the environment where the air conditioning control system, especially the CSI acquisition module, is located, without active interference from users or any large objects. The first wake-up threshold can be understood as the discrete statistical characteristics of the amplitude data of the channel state information in the resting state. The first wake-up threshold can be calculated from the amplitude data of the channel state information in the resting state. In this embodiment, the first wake-up threshold is predetermined.

[0037] In one embodiment, the next first preset duration can be continuous with the current first preset duration, thus each first preset duration determines a first discrete statistical feature. After determining the first discrete statistical feature corresponding to each first preset duration, the first discrete statistical feature is compared with a first wake-up threshold. If the first discrete statistical feature is less than the first wake-up threshold, the acquisition of amplitude data for the next first preset duration continues. The next first preset duration can be continuous with the previous first preset duration or separated by a first preset time interval. The time interval between two adjacent first preset durations is not limited here and can be set as needed. If the first discrete statistical feature is greater than or equal to the first wake-up threshold, the amplitude data of the second channel state information within the second preset duration is acquired.

[0038] The first moment is before the start of the current first preset duration, and the second moment is after the start of the current second preset duration. Therefore, the duration between the first and second moments, i.e., the second preset duration, includes the first preset duration. If only the CSI data from the first preset duration onwards is considered, the initial stage of the action might be missed; if only the CSI data from the first preset duration onwards is considered, the subsequent stages of the action might be missed. Therefore, setting the second preset duration to the duration between the first and second moments ensures that the second channel state information includes the first channel state information, guaranteeing the accuracy of the CSI data. Consequently, the amplitude data of the second channel state information can more accurately reflect the action characteristics, thus more accurately determining whether to activate the air conditioner's gesture recognition function.

[0039] In the technical solution disclosed herein, before activating the gesture recognition function of the air conditioner, the control system of the air conditioner is in standby mode, and in standby mode, steps S11 to S13 are executed.

[0040] In related technologies, air conditioners with gesture recognition continuously perform highly complex feature extraction and complex model matching on the CSI data stream, causing the Wi-Fi module's main processor to run for extended periods, resulting in excessively high overall energy consumption of the air conditioning control system. Furthermore, there is a high rate of false triggering and misrecognition, potentially misinterpreting unintentional user actions (such as getting up, turning around, or pet activity) as control gestures, leading to frequent system malfunctions and severely impacting user experience.

[0041] The technical solution disclosed herein establishes a standby mode before activating the gesture recognition function of the air conditioner. In standby mode, amplitude data of multiple first channel state information within a first preset time period are acquired, and a first discrete statistical characteristic of the amplitude data of the multiple first channel state information is determined. In response to the current first discrete statistical characteristic being greater than or equal to a first wake-up threshold, amplitude data of second channel state information within a second preset time period is acquired. Based on the amplitude data of the second channel state information, it is determined whether to activate the gesture recognition function of the air conditioner. Therefore, before activating the gesture recognition function of the air conditioner, i.e., in standby mode, the amplitude data of multiple first channel state information within the first preset time period is processed every first preset time interval. Processing of the amplitude data of the second channel state information within the second preset time period is only required when the current first discrete statistical characteristic is greater than or equal to the first wake-up threshold. This method eliminates the need for continuous high-intensity extraction of the CSI data stream, significantly reducing the power consumption of the air conditioner control system and achieving a low-power standby mode.

[0042] Furthermore, the technical solution disclosed herein includes a first wake-up threshold. Only when the current first discrete statistical characteristic is greater than the first wake-up threshold is it necessary to acquire the amplitude data of the second channel state information within a second preset time period to determine whether to wake up the gesture recognition function. In other words, before waking up the gesture recognition function, a two-level judgment rule is set: not only must the current first discrete statistical characteristic be greater than or equal to the first wake-up threshold, but the amplitude data of the second channel state information must also be used to determine whether to wake up the gesture recognition function. This approach reduces false wake-up caused by unconscious user actions, improves trigger recognition accuracy, reduces the false trigger rate, effectively eliminates transient environmental interference, and improves the robustness of the control system.

[0043] In one embodiment, the first preset duration includes N consecutive time slices, and the first discrete statistical feature includes the median absolute deviation. Acquiring amplitude data of multiple first channel state information within the first preset duration and determining the first discrete statistical feature of the amplitude data of the multiple first channel state information may include: for each of the N consecutive time slices, acquiring p amplitude data of the first channel state information within each time slice to construct an amplitude data group, the amplitude data group including N*p amplitude data; determining the median of the amplitude data group; and determining the median absolute deviation based on the median and the amplitude data of each first channel state information.

[0044] A "time slice" can be understood as a "time window." The first preset duration can be divided into N consecutive time slices, or in other words, N time slices constitute the first preset duration. Within each time slice, p amplitude data points of the first channel state information are acquired. For example, with 10 time slices, each time slice being 50ms, and one amplitude data point acquired every 5ms, then 50 / 5 = 10 amplitude data points can be acquired in one time slice. Therefore, a total of 100 amplitude data points can be acquired over 10 time slices. These 100 amplitude data points can be constructed into an amplitude data set A = {A_1, A_2, ..., A_100}. The median of this amplitude data set is determined, and then, based on the median and each amplitude data point, the median absolute deviation is determined. D MAD1 , D MAD1 This can be expressed as formula (1): Formula (1) in, median(a) This represents the median of the amplitude data set. A i This represents each amplitude data point.

[0045] It is understandable that the first statistical feature of dispersion is not limited to the median absolute deviation of the amplitude data group A. Other mathematical parameters, such as the interquartile range, can also be used as the first statistical feature of dispersion. For example, the interquartile range of the amplitude data group A can be obtained by calculating the interquartile range and used as the first statistical feature of dispersion.

[0046] In this embodiment, the first discrete statistical feature is the median absolute deviation. The median absolute deviation is highly robust to sudden spike noise, which can effectively eliminate instantaneous environmental interference and further improve the robustness of the system.

[0047] The first wake-up threshold τ1 is determined based on background noise statistics of the system in a resting state. In one embodiment, the first wake-up threshold is calculated based on the mean and standard deviation of a discrete statistical feature sequence, which includes multiple second discrete statistical features, each of which is calculated from multiple amplitude data from channel state information in a resting state.

[0048] For example, multiple amplitude data points are acquired from a CSI data stream in a resting state, forming amplitude data groups. Each amplitude data group contains multiple amplitude data points. A second discrete statistical feature is calculated for each amplitude data group, resulting in a discrete statistical feature sequence. The mean and standard deviation of this discrete statistical feature sequence are calculated, and a first wake-up threshold is determined based on these mean and standard deviation.

[0049] In one embodiment, the second discrete statistical feature can be the median absolute deviation. D MAD2 The first wake-up threshold is denoted as τ1, and the first wake-up threshold τ1 can be expressed as formula (2): =μNoise+ k σNoise formula (2) Where μNoise is the data collected by the system in its resting state. D MAD2 The mean of the sequence, σNoise is D MAD2 The standard deviation of the sequence, k is a preset multiplication factor (e.g., k≥3), used to ensure that the false wake-up rate of the system is lower than the preset value when there is no user activity.

[0050] In one embodiment, the gesture recognition control method for the air conditioner can be executed by the gesture recognition control system of the air conditioner. The system may include a first processor LPP and a second processor MPP. The first processor LPP can execute step S11. After determining the first discrete statistical feature, the first processor LPP will... D MAD1 Compare with the first wake-up threshold τ1, when D MAD1 When the threshold is >τ1, the first processor LPP sends a first wake-up signal to the second processor MPP to wake up the second processor MPP. After receiving the first wake-up signal, the second processor MPP enters the trigger gesture preprocessing mode. In the trigger gesture preprocessing mode, it acquires the amplitude data of the second channel state information within a second preset time period. The time interval for acquiring the amplitude data within the second preset time period can be set as needed and is not specifically limited here.

[0051] In one embodiment, determining whether to wake up the gesture recognition function of the air conditioner based on the amplitude data of the second channel state information includes: determining the spectrum information of the second channel state information based on the amplitude data of the second channel state information; and determining whether to wake up the gesture recognition function of the air conditioner based on the spectrum information.

[0052] For example, the amplitude data of the second channel state information can be subjected to a short-time Fourier transform (STFT) to obtain the spectral information. Conventional techniques in the art can be used to perform the STFT on the amplitude data of the second channel state information; the specific process will not be detailed here.

[0053] It is understandable that other transformation methods can be used to obtain spectral information, and it is not limited to the short-time Fourier transform.

[0054] In one embodiment, determining whether to activate the air conditioner's gesture recognition function based on spectrum information may include: determining the total spectral energy within a preset trigger frequency range, where the preset trigger frequency range characterizes the action frequency range to which the air conditioner's trigger gesture belongs; and determining whether to activate the air conditioner's gesture recognition function based on the total spectral energy. For example, the process of determining the total spectral energy can be called periodic feature extraction.

[0055] The trigger gestures for an air conditioner can be understood as gestures that can trigger the air conditioner to adjust its state. These trigger gestures can be pre-set periodic trigger actions, such as rapid, repetitive arm waving, or circling movements.

[0056] The preset trigger frequency range is denoted as [ f low ,f high The rapid, repetitive swinging frequency of the human arm typically falls between 1Hz and 3Hz; therefore, a preset trigger frequency range can be used. f low ,f high The frequency range is set to [1Hz, 3Hz]. Based on the spectrum information, the total spectral energy within the preset trigger frequency range is calculated. E band This will serve as a periodic characteristic for determining whether it constitutes a "two-wave rapid hand gesture". Total spectral energy E band The calculation formula is as shown in formula (3): Formula (3) in, f τ represents frequency, and τ represents time. Indicates time τ and frequency f Energy density at that location.

[0057] In one embodiment, determining whether to wake up the gesture recognition function of the air conditioner based on the total spectrum energy may include: obtaining the matching confidence of the total spectrum energy and the trigger gesture of the air conditioner; and waking up the gesture recognition function of the air conditioner in response to the matching confidence being greater than a second wake-up threshold.

[0058] Determine the total spectral energy E band Then, the total spectral energy can be obtained. E band The confidence level of matching the trigger gesture with the air conditioner. The total spectral energy can be determined using a preset determination method. E band Confidence level of matching with the trigger gesture of the air conditioner.

[0059] For example, the total spectral energy can be input into a pre-trained binary classification model for trigger gestures. M trigger This yields the matching confidence score. For example, the total spectral energy can be input into a pre-trained binary classification model for trigger gestures. M trigger Triggering a gesture binary classification model M trigger It can output the matching confidence level.

[0060] Gesture-based binary classification model M trigger It can be obtained after training a pre-defined neural network model. (Trigger gesture binary classification model) M trigger Used to determine the total spectral energy of the input E band Which category does it belong to: "triggered" or "non-triggered" gesture?

[0061] For example, a pre-defined neural network model can be trained using multiple gesture samples and the corresponding confidence samples for each gesture sample to obtain a trigger gesture binary classification model. M trigger .

[0062] The second wake-up threshold τ2 is preset. The second wake-up threshold τ2 can also be called the confidence threshold. It can be determined in the following way to determine whether to wake up the gesture recognition function.

[0063] The determination of the second wake-up threshold τ2 mainly follows the F1-Score optimization principle. The F1-Score is a statistical metric for measuring the accuracy of a binary classification model. It is the harmonic mean of precision and recall.

[0064] Precision: The proportion of samples that are actually trigger gestures among all samples classified as "trigger gestures" (positive class) by the binary classification model. The higher the accuracy, the lower the false wake-up rate of the control system (i.e., the trigger gestures that wake up the control system are rarely erroneous non-trigger gestures).

[0065]

[0066] Recall (Recall Rate): The recall rate is the proportion of samples that are actually "trigger gestures" (positive class) that are correctly identified as trigger gestures by the binary classification model. The higher the recall rate, the lower the false negative rate of the system (i.e., the higher the probability that the control system can successfully recognize a user's trigger gesture).

[0067]

[0068] The calculation steps for the second wake-up threshold τ2 are as follows: (1) Use a pre-trained trigger gesture binary classification model M trigger Test the positive and negative sample validation sets of the triggered gestures (i.e., the data set containing real triggered gestures and non-triggered gestures).

[0069] (2) In the trigger gesture binary classification model M trigger The output confidence (typically between 0 and 1) is used to iterate through all possible values ​​as candidate thresholds τ_candidate.

[0070] (3) For each candidate threshold τ_candidate: All samples whose model output confidence >= τ_candidate are classified as "trigger gesture" (positive class).

[0071] All samples whose model output Confidence < τ_candidate are classified as "non-triggered gestures" (negative class).

[0072] Based on this determination, calculate the precision and recall for the current τ_candidate.

[0073] Substitute Precision and Recall into the F1-Score formula to calculate the current F1-Score value.

[0074] (4) After traversing all candidate thresholds τ_candidate, the Confidence value corresponding to the maximum value of F1-Score is selected as the final second wake-up threshold τ2.

[0075] (5) The second wake-up threshold τ2 can be embedded into the program of the air conditioner control system as a confidence threshold for determining whether to wake up the gesture recognition function of the air conditioner.

[0076] After obtaining the matching confidence score, the matching confidence score is compared with the second wake-up threshold. If the matching confidence score is greater than the second wake-up threshold, the gesture recognition function of the air conditioner is woken up. If the matching confidence score is less than or equal to the second wake-up threshold, the process returns to step S11 to continue obtaining the amplitude data of multiple first channel state information within the next first preset time period.

[0077] In this embodiment, the total spectral energy within a preset trigger frequency range is extracted, and this total spectral energy is used to match periodic triggering actions (such as two rapid hand waves). This can effectively distinguish non-periodic false actions such as single hand waves, walking, and environmental vibrations, thereby reducing the false trigger rate.

[0078] After the gesture recognition function of the air conditioner is activated, the control method may further include: performing gesture feature recognition on the second channel state information within a second preset time period to obtain gesture feature information; determining control commands based on the gesture feature information; and adjusting the operating state of the air conditioner based on the control commands.

[0079] If the matching confidence score is greater than the second wake-up threshold, it indicates that the gesture in the second channel state information is likely a trigger gesture of the air conditioner. Therefore, the gesture recognition function of the air conditioner is activated. After activating the gesture recognition function, gesture feature recognition can be performed on the data stream of the second channel state information within a second preset time period. This involves performing highly complex feature extraction (such as complete time-domain, frequency-domain, or subcarrier fusion processing) and complex model matching on the data stream of the second channel state information within the second preset time period to obtain gesture feature information. Control commands are determined based on the gesture feature information. The air conditioner's command execution module adjusts the operating state of the air conditioner according to the control commands.

[0080] This disclosure also provides a gesture recognition control device (also called a control system) for an air conditioner, which includes a first processor (LPP) and a second processor (MPP). The first processor is configured to acquire amplitude data of multiple first channel state information within a first preset time interval at first preset time intervals, and determine a first discrete statistical feature of the amplitude data of the multiple first channel state information. The first discrete statistical feature includes at least one of median absolute deviation and interquartile range. Exemplarily, the control system may further include a channel state information acquisition module (i.e., a CSI acquisition module), through which the first processor can acquire amplitude data of multiple first channel state information within the first preset time interval.

[0081] The second processor is used to acquire amplitude data of the second channel state information within a second preset duration when the current first discrete statistical feature is greater than the first wake-up threshold. The second preset duration is the duration between the first moment and the second moment, where the first moment is before the start time of the current first preset duration and the second moment is after the start time of the current second preset duration.

[0082] The second processor is also used to determine whether to wake up the gesture recognition function of the air conditioner based on the amplitude data of the second channel state information.

[0083] In one embodiment, the second processor is further configured to: perform gesture feature recognition on the second channel state information within a second preset time period to obtain gesture feature information; and determine a control command based on the gesture feature information. The device also includes an instruction execution module, which is configured to adjust the operating state of the air conditioner according to the control command.

[0084] The technical solution of this disclosure will be described in detail below through a specific embodiment. Figure 2 This is a schematic diagram illustrating the workflow of the gesture recognition control device of this disclosure. It should be noted that the specific values ​​in the following embodiments are only set for the convenience of illustrating the scheme and do not constitute a limitation on the specific scheme of this disclosure.

[0085] The control device is in a low-power standby mode. In this mode, every first preset time interval, the first processor acquires amplitude data of multiple first channel state information within a first preset time period through the CSI acquisition module. The first processor calculates the median absolute deviation.

[0086] In the following embodiments, it is assumed that the first wake-up threshold τ1 = 30, the trigger frequency range is f_min = 1 Hz, F_max = 3 Hz, and the second wake-up threshold τ2 = 0.90 determined by offline experiments.

[0087] For example, assuming one CSI amplitude data point is collected every 5 milliseconds, the duration of each time slice is T=50ms, and the number of consecutive time slices is N=10, then 50 / 5=10 amplitude data points can be collected within one time slice (50ms), and a total of 100 amplitude data points can be collected over 10 time slices. Calculate the median absolute deviation of the 100 amplitude data points.

[0088] Suppose that at a certain moment, a user's rapid hand-waving motion causes a change in the CSI amplitude signal.

[0089] Primary wake-up judgment (LPP execution): LPP collects the CSI amplitude data set A={A_1, A_2, ..., A_100} from the most recent N=10 time slices (a total of 100 amplitude data points) and calculates the median absolute deviation. D MAD1 .

[0090] Calculate the median: Assume the median CSI amplitude value of these 100 amplitude data points is 150. Calculate the median absolute deviation ( D MAD1 ): Calculate the absolute value of the difference between each amplitude data point and the median, and then take the median of these absolute values.

[0091] Assume the calculation result is: D MAD1 = median(|A_i - 150|) = 45 Comparison and Arousal: LPP will D MAD1 Compare with the first wake-up threshold τ1: D MAD1 = 45, τ1 = 30. Because 45 > 30, that is... D MAD1 When τ1 is established, LPP sends a wake-up signal to MPP, triggering MPP's trigger gesture preprocessing mode.

[0092] Periodic Feature Extraction (MPP Execution): Once the MPP is activated, it immediately acquires and analyzes the CSI amplitude data sequence A_stft (e.g., a total of 1024 amplitude data points, covering 5.12 seconds) within the second preset duration.

[0093] Short-time Fourier Transform: MPP performs a short-time Fourier transform on A_stft to obtain the signal's spectral information |STFT( , f)|.

[0094] Extracting the total spectral energy: The total spectral energy is extracted using formula (3) above. Assume the calculation result is... E band =25000, E band These are periodic features used to trigger the judgment of the gesture binary classification model.

[0095] Precise matching and state switching: MPP will Eband =25000 is input into the pre-trained trigger gesture binary classification model.

[0096] The trigger gesture binary classification model outputs the matching confidence score. Assume the trigger gesture binary classification model outputs: Confidence = 0.95.

[0097] The system compares Confidence with the second wake-up threshold τ2 = 0.90. Since Confidence > τ2, the control system recognizes the gesture as a trigger and then enters the air conditioner's full-function gesture recognition (i.e., gesture recognition function), performs command mapping, and executes the corresponding control command. If Confidence < τ2, the control system immediately returns to low-power standby mode.

[0098] The technical solution disclosed herein operates with low power consumption in low-power standby mode, preventing excessive energy consumption caused by continuous high-intensity CSI data detection in the control system. This keeps the second processor in sleep or low-frequency state most of the time, significantly reducing system energy consumption compared to CSI gesture recognition solutions in related technologies. Furthermore, by setting two levels of judgment rules, it effectively distinguishes between unconscious user actions and preset trigger gestures, reducing the false trigger rate, effectively eliminating transient environmental interference, and improving the robustness of the control system. This technical solution only requires the use of the air conditioner's built-in Wi-Fi module, incurring no additional hardware costs and avoiding the increased costs and privacy issues associated with adding cameras or voice modules.

[0099] According to embodiments of this disclosure, an electronic device is also provided. The electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to perform the control method of any embodiment of this disclosure.

[0100] This disclosure also provides a readable storage medium storing a computer program that, when executed by a processor, implements the control method as described in any embodiment of this disclosure.

[0101] One embodiment of this disclosure also provides an air conditioner, which includes the control system of this disclosure embodiment, or the electronic device of this disclosure embodiment.

[0102] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0103] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0104] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0105] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0106] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0107] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0108] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0109] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this disclosure, "multiple" means two or more, unless otherwise explicitly specified.

[0110] The foregoing disclosure provides many different implementations or examples for carrying out different structures of this disclosure. To simplify this disclosure, the components and arrangements of specific examples are described above. Of course, these are merely examples and are not intended to limit this disclosure. Furthermore, reference numerals and / or reference letters may be repeated in different examples; such repetition is for simplification and clarity and does not in itself indicate a relationship between the various implementations and / or arrangements discussed.

[0111] The above are merely specific embodiments of this disclosure, but the scope of protection of this disclosure is not limited thereto. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure, and any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed herein, and the combination of different parts of different embodiments without conflict, should all be covered within the scope of protection of this disclosure. Therefore, the scope of protection of this disclosure should be determined by the scope of the claims.

Claims

1. A gesture recognition control method of an air conditioner, characterized by, Comprising: S11: obtaining amplitude data of a plurality of first channel state information within a first preset time length, and determining a first dispersion statistical feature of the amplitude data of the plurality of first channel state information, the first dispersion statistical feature including at least one of median absolute deviation and interquartile range; S12: in response to the current first dispersion statistical feature being greater than or equal to a first wake-up threshold, obtaining amplitude data of second channel state information within a second preset time length, the second preset time length being a time length between a first time and a second time, the first time being located before a starting time of the current first preset time length, the second time being located after a starting time of the current second preset time length, the first wake-up threshold representing a dispersion statistical feature of amplitude data of channel state information in a resting state; S13: determining whether to wake up a gesture recognition function of the air conditioner according to the amplitude data of the second channel state information.

2. The method of claim 1, wherein, The first preset time length includes N consecutive time slices, and the first dispersion statistical feature includes median absolute deviation. The step of obtaining amplitude data of a plurality of first channel state information within a first preset time length, and determining a first dispersion statistical feature of the amplitude data of the plurality of first channel state information, comprises: For the N consecutive time slices, respectively obtaining amplitude data of p first channel state information in each of the time slices to construct an amplitude data group, the amplitude data group including N*p amplitude data; Determining the median of the amplitude data group; Determining the median absolute deviation according to the median and the amplitude data of each of the first channel state information.

3. The method of claim 1, wherein, Further comprising: The first wake-up threshold is calculated according to an average value and a standard deviation of a dispersion statistical feature sequence, the dispersion statistical feature sequence including a plurality of second dispersion statistical features, each of the second dispersion statistical features being calculated from a plurality of amplitude data in channel state information in a resting state.

4. The method according to any one of claims 1-3, characterized in that, The step of determining whether to wake up a gesture recognition function of the air conditioner according to the amplitude data of the second channel state information, comprises: Determining spectral information of the second channel state information according to the amplitude data of the second channel state information; Determining whether to wake up the gesture recognition function of the air conditioner according to the spectral information.

5. The method of claim 4, wherein, The step of determining spectral information of the second channel state information according to the amplitude data of the second channel state information, comprises: Performing short-time Fourier transform on the amplitude data of the second channel state information to obtain the spectral information.

6. The method of claim 4, wherein, The step of determining whether to wake up the gesture recognition function of the air conditioner according to the spectral information, comprises: Determining total spectral energy in a preset trigger frequency range according to the spectral information, the preset trigger frequency range being used to represent a motion frequency range to which a trigger gesture of the air conditioner belongs; Determining whether to wake up the gesture recognition function of the air conditioner according to the total spectral energy.

7. The method of claim 6, wherein, The step of determining whether to wake up the gesture recognition function of the air conditioner according to the total spectral energy, comprises: Obtaining a matching confidence of the total spectral energy and the trigger gesture of the air conditioner; In response to the matching confidence level being greater than the second wake-up threshold, the gesture recognition function of the air conditioner is activated.

8. The method of claim 7, wherein, Also includes: In response to the matching confidence being less than or equal to the second wake-up threshold, the process returns to step S11 to obtain amplitude data of multiple first channel state information within the next first preset time period, and to determine the first discrete statistical feature of the amplitude data of the multiple first channel state information.

9. The method of claim 7, wherein, The step of obtaining the matching confidence level between the total spectral energy and the trigger gesture of the air conditioner includes: The total spectral energy is input into a pre-trained trigger gesture binary classification model to obtain the matching confidence. The trigger gesture binary classification model is trained using multiple gesture samples and the confidence samples corresponding to each gesture sample.

10. The method of claim 7, wherein, Also includes: Gesture feature recognition is performed on the second channel state information within the second preset time period to obtain gesture feature information; Based on the gesture feature information, the control command is determined; The operating status of the air conditioner is adjusted according to the control command.

11. The method of claim 1, wherein, Also includes: In response to the fact that the current first discrete statistical feature is less than the first wake-up threshold, return to step S11, obtain the amplitude data of multiple first channel state information within the next first preset time period, and determine the first discrete statistical feature of the amplitude data of the multiple first channel state information.

12. A gesture recognition control apparatus of an air conditioner, characterized by, include: A first processor is configured to acquire amplitude data of multiple first channel state information within a first preset time period, and determine a first discrete statistical feature of the amplitude data of the multiple first channel state information, wherein the first discrete statistical feature includes at least one of median absolute deviation and interquartile range. The second processor is configured to acquire amplitude data of the second channel state information within a second preset duration when the current first discrete statistical feature is greater than the first wake-up threshold. The second preset duration is the duration between the first moment and the second moment, where the first moment is before the start time of the current first preset duration and the second moment is after the start time of the current second preset duration. The second processor is further configured to determine whether to activate the gesture recognition function of the air conditioner based on the amplitude data of the second channel state information.

13. The apparatus according to claim 12, characterized in that, The second processor is further configured to: perform gesture feature recognition on the second channel state information within the second preset time period to obtain gesture feature information; and determine control commands based on the gesture feature information; The device further includes an instruction execution module, which is used to adjust the operating state of the air conditioner according to the control instructions.

14. An electronic device, comprising: include: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-11.

15. An air conditioner characterized by comprising: It includes the control device as described in claim 12 or 13, or the electronic device as described in claim 14.

Images