A vehicle trunk control method and system

Abstract

The application provides a vehicle trunk control method and system, and relates to the technical field of intelligent vehicle control. The millimeter wave radar arranged at the tail of the vehicle is activated in response to the opening of the vehicle trunk. The left and right swinging actions of the user's legs are identified through point cloud clustering and micro-Doppler spectrum analysis based on the echo data collected by the millimeter wave radar. In response to the left and right swinging actions of the legs, the vehicle trunk is controlled to be closed, so as to improve the convenience, identification accuracy and anti-interference ability of the trunk control.

Description

Technical Field

[0001] This invention relates to the field of vehicle intelligent control technology, and more specifically, to a vehicle trunk control method and system. Background Technology

[0002] As automobiles become increasingly intelligent, the sensor-controlled function of electric tailgates is becoming more and more common. In existing technologies, non-contact tailgate control solutions generally adopt a foot-kicking action mode, which involves installing a capacitive sensor or millimeter-wave radar under the rear bumper of the vehicle. The user needs to extend their foot into a specific area under the bumper and make a kicking motion. The tailgate is opened or closed by detecting the signal change generated by the foot movement.

[0003] However, this kick-based control method has significant shortcomings in practical use: users need to stand on one foot and accurately insert their foot under the bumper, which is difficult and poses safety hazards when carrying items with both hands, on slippery ground, or when mobility is limited; the kicking action is easily interfered with by external factors such as small animals running around, clothing swinging, and road debris, resulting in a high false trigger rate. In addition, existing millimeter-wave radar kick sensing solutions mainly focus on simple matching and recognition of foot point clouds or trajectories, making it difficult to effectively distinguish between genuine operation intentions and accidental disturbances, and both user experience and reliability need to be improved. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a vehicle trunk control method and system to improve the convenience, identification accuracy and anti-interference ability of trunk control.

[0005] Firstly, this application provides a method for controlling the trunk of a vehicle, including: The millimeter-wave radar located at the rear of the vehicle is activated in response to the opening of the vehicle's trunk. Based on the echo data collected by millimeter-wave radar, the user's left and right leg swinging motion is identified by point cloud clustering and micro-Doppler spectrum analysis. The left and right leg swinging motion is the reciprocating swinging of the user's legs in the horizontal direction. The system responds to the left and right swinging motion of the legs to control the closing of the vehicle's trunk.

[0006] Optionally, the echo data is point cloud data; based on the echo data acquired by millimeter-wave radar, the user's left-right leg swinging motion is identified through point cloud clustering and micro-Doppler spectrum analysis, including: Clustering of point cloud data yields point cloud clusters; Based on the vertical coordinate information in the point cloud data, the lower limb dynamic point cloud cluster is separated from the point cloud cluster; Determine the horizontal displacement trajectory of the centroid of the lower limb dynamic point cloud cluster, and obtain the user's micro-Doppler spectrum based on the echo data corresponding to the lower limb dynamic point cloud cluster; When the horizontal displacement trajectory of the center of mass has a preset number of alternating left and right swing cycles within a preset time period and the swing amplitude is greater than a preset amplitude threshold, and the matching degree between the micro-Doppler spectrum and the preset leg swing template exceeds a preset matching threshold, the user's left and right leg swing motion is determined to be detected.

[0007] Optionally, before identifying the user's left-right leg swinging motion based on the echo data collected by millimeter-wave radar through point cloud clustering and micro-Doppler spectrum analysis, the following steps are also included: Based on distance information in the echo data, determine the distance between the user and the vehicle; When the distance is less than a preset distance threshold and the absolute value of the user's radial velocity relative to the vehicle is less than a preset speed threshold, the user's leg swinging motion is recognized.

[0008] Optionally, the horizontal displacement trajectory of the centroid of the lower limb dynamic point cloud cluster is determined, including: Determine the centroid coordinates of the lower limb dynamic point cloud clusters in the horizontal direction in each frame to form a trajectory sequence; The trajectory sequence is bandpass filtered, and peak and valley values ​​are detected in the filtered trajectory sequence; where an effective leg swinging motion cycle includes a local maximum and a local minimum value of the horizontal displacement trajectory.

[0009] Optionally, the matching degree between the micro-Doppler spectrum and the preset leg swing template exceeds a preset matching threshold, including: Extract the spectral features of the micro-Doppler spectrum and check whether the spectral features exhibit symmetrical and periodic upper and lower sideband characteristics; The matching degree between the spectral features and the standard spectral features in the preset leg swing template is calculated. When the matching degree exceeds the preset matching threshold, it is determined that the frequency domain condition is met.

[0010] Optionally, controlling the closing of the vehicle's trunk in response to a left-right leg swing includes: Based on the left and right swinging motion of the legs, a command to close the trunk is generated; The command to close the trunk is encapsulated into a CAN data frame and sent to the trunk control module. Before closing the trunk, the trunk control module performs a safety check to confirm that the trunk is open and the vehicle is in park or the vehicle speed is below a preset safety threshold. During the closing process, an anti-pinch detection is performed. If an abnormal increase in resistance is detected, the closing process stops and the device is opened in reverse.

[0011] Optionally, after activating the millimeter-wave radar located at the rear of the vehicle, the following are also included: The indicator lights located inside the trunk or at the rear of the vehicle enter a guidance state to guide the user to the interactive position directly behind the rear of the vehicle.

[0012] Secondly, this application provides a vehicle trunk control system, including: The status detection module is used to activate the millimeter-wave radar located at the rear of the vehicle in response to the opening of the vehicle's trunk. The data processing module is used to identify the user's left and right leg swinging motion based on the echo data collected by millimeter-wave radar through point cloud clustering and micro-Doppler spectrum analysis. The left and right leg swinging motion is the reciprocating swinging of the user's legs in the horizontal direction. The control execution module is used to control the closing of the vehicle's trunk in response to the left and right swinging motion of the legs.

[0013] Thirdly, this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described vehicle trunk control method.

[0014] Fourthly, this application provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement the aforementioned vehicle trunk control method.

[0015] This application provides a vehicle trunk control method and system. In response to the opening of the vehicle trunk, a millimeter-wave radar located at the rear of the vehicle is activated; echo data obtained from the millimeter-wave radar scan is acquired; based on the echo data, the user's left-right leg swinging motion is identified through point cloud clustering and micro-Doppler spectrum analysis; in response to the identified left-right leg swinging motion, a trunk closing command is generated and sent to the trunk control module to drive the electric mechanism to close the trunk, thereby improving the convenience, accuracy, and anti-interference capability of trunk control.

[0016] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart of a vehicle trunk control method provided by an embodiment of the present invention is shown; Figure 2 This diagram illustrates the structure of a vehicle trunk control system provided in an embodiment of the present invention. Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present invention is shown. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0020] The vehicle trunk control method provided in this application is applied to a vehicle trunk control system. The control system includes a trunk status detection module, a millimeter-wave radar module, a body control module, a trunk control module, indicator lights, and an audio output device. The system includes a trunk status detection module that detects the trunk latch status via a microswitch or Hall sensor within the lock body and sends the status signal to the body control module. A millimeter-wave radar module, including a millimeter-wave radar and signal processing circuitry, is installed at the rear of the vehicle. It scans the rear area and outputs the signal to the signal processing circuitry, which processes the echo data to identify the user's leg movements and sends the results to the body control module. The body control module (BCM) is connected to both the millimeter-wave radar module and the trunk control module via a CAN bus to generate and send closing commands. The trunk control module is connected to an electric strut or drive motor located at the connection between the trunk lid and the body, driving the strut or motor to rotate the trunk lid around the hinge to close it. An indicator light is connected to the body control module, which controls the indicator light to enter a guidance state to guide user interaction. An audio output device is connected to the body control module to emit prompts during the guidance process to encourage user interaction.

[0021] This application provides a method for controlling a vehicle trunk, see below. Figure 1 As shown, the vehicle trunk control method provided in this application embodiment includes at least the following steps: Step 110: In response to the opening of the vehicle's trunk, activate the millimeter-wave radar located at the rear of the vehicle.

[0022] In this embodiment, when the vehicle trunk is opened, the trunk status detection module detects the trunk latch changing from locked to released state via a microswitch or Hall sensor inside the lock body, determining that the trunk has been physically opened. Subsequently, the body control module or gateway sends a wake-up signal to the millimeter-wave radar module installed at the rear of the vehicle via the CAN bus or puts the millimeter-wave radar module into a high-power operating mode. After power-on, the millimeter-wave radar module performs a rapid self-test to confirm that the transmitter, receiver, and signal processing unit of the millimeter-wave radar module are functioning normally, preparing for subsequent continuous scanning.

[0023] In this embodiment, the millimeter-wave radar is positioned at the rear of the vehicle. Compared to existing foot-kicking solutions where sensors are typically mounted below the rear bumper, with their detection beam angled downwards and pointing towards the narrow gap between the lower bumper and the ground to capture foot insertion and kicking motions, this embodiment places the millimeter-wave radar near the taillights or next to the trunk lid, at a height of approximately 0.55 to 0.7 meters above the ground. The detection beam is horizontally directed towards the rear of the vehicle, focusing on covering the legroom of a user standing behind the vehicle. This adjustment in installation position is adapted to the new interactive motion mode of leg swinging, allowing the radar to directly scan the user's leg area, obtaining more complete and stable leg point cloud data, laying the hardware foundation for subsequent accurate motion recognition.

[0024] To guide users to quickly find the optimal interaction position and reduce operational difficulty, in this embodiment, after activating the millimeter-wave radar located at the rear of the vehicle, an indicator light located inside the trunk or at the rear of the vehicle is controlled to enter a guidance state to guide the user to the interaction position directly behind the rear of the vehicle. When the millimeter-wave radar module is activated, if a potential target is detected within a wide detection area, the vehicle control module controls the indicator light to enter a standby guidance state with slow, breathing flashes, indicating to the user that the vehicle control module is ready. If the user does not enter the effective area, the indicator light may remain off or not enter the guidance state, silently prompting the user to continue adjusting the position. As a more preferred embodiment, the indicator light can be a multi-area LED light strip, combined with a short prompt sound emitted by an audio output device, actively guiding the user to the optimal interaction position directly behind the rear of the vehicle through light flow or directional light spots.

[0025] Specifically, the millimeter-wave radar is preferably installed near the taillights or next to the trunk door, at a height of about 0.55 meters to 0.7 meters above the ground. The effective detection angle of the millimeter-wave radar covers the main area behind the trunk opening, and is specially optimized to cover the area around the pedestrian's legs.

[0026] Regarding radar installation, existing foot-kicking solutions typically mount sensors below the rear bumper, with their detection beam angled downwards and pointing towards the narrow gap between the lower side of the bumper and the ground, aiming to capture the action of extending and kicking the foot. In this embodiment, however, the millimeter-wave radar is installed near the taillights or next to the trunk lid, at a height of approximately 0.55 to 0.7 meters above the ground. Its detection beam is horizontally directed towards the rear of the vehicle, focusing on covering the legroom of a user standing behind the vehicle. This adjustment in installation position is adapted to the new interactive action mode of leg swinging from side to side, allowing the radar to directly scan the user's leg area, obtaining more complete and stable leg point cloud data, laying the hardware foundation for subsequent accurate action recognition.

[0027] It should be noted that in this embodiment, the user does not need to extend their foot under the rear bumper to kick as in existing foot-kicking sensing solutions. Instead, they only need to stand directly behind the rear of the vehicle and issue the closing command by swinging their leg back and forth horizontally. Compared to the kicking action, this interaction method has a fundamental difference in action mode: First, the user does not need to stand on one leg; both legs remain firmly on the ground, making operation safer and more natural in complex scenarios such as carrying items with both hands, slippery ground, or difficulty in movement. Second, the interaction area has shifted from the narrow space under the bumper to the open area behind the rear of the vehicle. The user does not need to precisely find the kicking position; they only need to stand about 0.5 meters directly behind the rear of the vehicle according to the light guidance to complete the operation, significantly reducing the difficulty of operation and the threshold for use.

[0028] In one optional embodiment, the control system further includes an AR projection module and an onboard camera. The AR projection module is located at the rear of the vehicle and is used to project an interactive area and guiding animation onto the ground; the onboard camera is used to acquire images of the rear of the vehicle for visual confirmation. The AR projection module is preferably a DLP micro-projection module, installed below the inside of the trunk lid or above the rear bumper, with the projection window facing downwards and towards the ground. The AR projection module is connected to the body control module via an onboard Ethernet or CAN bus, and the projected content is rendered in real time by the body control module or a dedicated projection control unit.

[0029] Specifically, AR projection guidance is used in conjunction with a vision-radar multimodal fusion interaction method. The vehicle control module dynamically generates projection content based on the user's position, direction of movement, and distance detected by the millimeter-wave radar module, and projects this content onto the ground behind the vehicle via the AR projection module. The projection content includes a dynamic interactive area frame, swaying guidance animations, or directional arrows. The vehicle control module dynamically adjusts the projection position and direction based on the user's position detected by the radar, ensuring the projection always falls within the user's field of vision. For example, when the user is located at the left rear of the vehicle, the projected pattern is tilted to the left; as the user moves, the projected pattern moves synchronously to maintain the visibility of the guidance information.

[0030] Furthermore, while performing radar identification, if the vehicle is equipped with a rearview or surround-view camera, the vehicle control module or central computing platform simultaneously acquires camera images to visually confirm the target within the projection area. Visual confirmation uses a human pose estimation algorithm to identify the user's leg position and posture, and performs spatiotemporal registration and cross-validation with the radar point cloud data. Specifically, the dynamic lower limb point cloud clusters in the radar point cloud are mapped to the image coordinate system, and their overlap with the key points or bounding boxes of the human legs detected by the camera is calculated. Only when both the radar identification result and the visual confirmation result simultaneously meet preset conditions does the subsequent action recognition process begin. This visual-radar multimodal fusion verification mechanism effectively eliminates false detections that may occur when radar perceives alone, further improving system reliability.

[0031] Furthermore, the projected pattern changes as the user approaches: when the user is at a distance, the projection displays directional indicators or prompts to move closer; when the user enters a preset close-range area, the projection displays a demonstration animation of the user's legs swinging left and right or a light trail, intuitively guiding the user to complete the prescribed actions. This AR projection guidance method transforms the passive waiting for the user to discover the interaction location into actively projecting guidance information into the user's field of vision, which is especially suitable for scenarios at night, in rainy weather, or where there are no obvious markings on the ground, significantly improving interaction efficiency.

[0032] In this embodiment of the application, the millimeter-wave radar transmits frequency-modulated continuous wave (FMCW) and accurately calculates the distance and relative speed to the target by calculating the time difference and frequency difference between the transmitted and received signals.

[0033] Specifically, millimeter-wave radar continuously scans the effective detection area (usually a fan-shaped area) and receives reflected echo data. The echo data contains information such as the target's distance, velocity, angle, and signal strength, providing the raw data foundation for subsequent point cloud generation and micro-Doppler analysis.

[0034] Furthermore, the signal processing circuit in the millimeter-wave radar module processes multiple consecutive frames of detection data. These consecutive frames are echo data output by the millimeter-wave radar over multiple consecutive scan cycles, with each frame corresponding to target distance, velocity, and angle information acquired within one scan cycle. The signal processing circuit first uses Moving Target Indication (MTI) technology to filter out stationary background clutter such as ground and walls. Then, by performing Doppler effect analysis on the remaining signal, it identifies moving targets with radial velocity. When a moving target is confirmed to exist within a preset critical distance of 0.5 meters, it is initially marked as a potential pedestrian target, and a more refined motion recognition algorithm is activated.

[0035] Step 120: Based on the echo data collected by millimeter-wave radar, identify the user's left and right leg swinging motion through point cloud clustering and micro-Doppler spectrum analysis. The left and right leg swinging motion is the reciprocating swinging of the user's legs in the horizontal direction.

[0036] In this embodiment, before recognizing the user's left-right leg swinging motion based on echo data, a context-aware decision is first performed: the distance between the user and the vehicle is determined based on the distance information in the echo data; the user's left-right leg swinging motion is only recognized when the distance is less than a preset distance threshold and the absolute value of the user's radial velocity relative to the vehicle is less than a preset speed threshold. The preset distance threshold is, for example, 0.5 meters, and the preset speed threshold is, for example, 0.1 meters per second. An absolute value of radial velocity less than the preset speed threshold indicates that the user is relatively stationary. Only when the above context conditions are met will subsequent periodic leg movements be recognized as a closing intention, thereby avoiding accidental triggering when pedestrians pass by.

[0037] After satisfying the above context-aware decision conditions, the following triple criterion fusion recognition process, consisting of time-domain trajectory analysis, frequency-domain feature verification, and context-aware decision, is executed: Point cloud data is clustered to obtain point cloud clusters; based on the vertical coordinate information in the point cloud data, the lower limb dynamic point cloud cluster is separated from the point cloud cluster; the centroid horizontal displacement trajectory of the lower limb dynamic point cloud cluster is determined, and the user's micro-Doppler spectrum is obtained based on the echo data corresponding to the lower limb dynamic point cloud cluster; when the centroid horizontal displacement trajectory has a preset number of alternating left and right swing cycles within a preset time period and the swing amplitude is greater than a preset amplitude threshold, and the matching degree between the micro-Doppler spectrum and the preset leg swing template exceeds a preset matching threshold, the user's left and right leg swinging motion is determined to be recognized.

[0038] Furthermore, the centroid horizontal displacement trajectory of the lower limb dynamic point cloud cluster is determined, specifically including: determining the centroid coordinates of the lower limb dynamic point cloud cluster in the horizontal direction in each frame to form a trajectory sequence; performing bandpass filtering on the trajectory sequence, and detecting peak and valley values ​​on the filtered trajectory sequence; wherein, an effective leg swinging motion cycle includes a local maximum value and a local minimum value of the horizontal displacement trajectory.

[0039] Determining whether the matching degree between the micro-Doppler spectrum and the preset leg swing template exceeds the preset matching threshold specifically includes: extracting the spectral features of the micro-Doppler spectrum and checking whether the spectral features exhibit symmetrical and periodic upper and lower sideband features; calculating the matching degree between the spectral features and the standard spectral features in the preset leg swing template; when the matching degree exceeds the preset matching threshold, it is determined that the frequency domain condition is met.

[0040] In one optional embodiment, after satisfying the context-aware decision condition, a dynamic environmental interference adaptive suppression mechanism based on Raney entropy is further implemented. Raney entropy, as a generalized form of information entropy, can characterize the dispersion of radar point cloud distribution in spatial or velocity dimensions. Specifically, for clustered point cloud data or lower limb dynamic point cloud clusters, the Raney entropy value of each frame of point cloud in the horizontal, vertical, or radial velocity dimensions is calculated, forming an entropy time series. The formula for calculating Raney entropy is:

[0041] In the formula, Let the entropy value be the value of the Rainier. For the order parameter, For the point cloud in the corresponding dimension The probability distribution of each component. It can take real numbers greater than 0 and not equal to 1, for example When the value is 2, the Raney entropy is more sensitive to high-probability regions and is suitable for detecting dynamically distributed targets.

[0042] Effective leg swinging motions produce regular, periodic fluctuations in entropy, with the fluctuation period matching the swing period and the entropy value variation amplitude relatively stable. In contrast, random interference such as wind blowing debris, a pet's tail wagging, or radar crosstalk from passing vehicles causes unpredictable entropy changes, or exhibits sudden, non-periodic fluctuations. The vehicle control module automatically adjusts the entropy detection threshold based on ambient background noise: in severe weather conditions such as strong winds, rain, or snow, the overall background entropy increases, and the detection threshold increases accordingly; in quiet environments, the detection threshold decreases accordingly. Only when the fluctuation pattern of the Reni entropy time series matches the preset periodic swing characteristics does it proceed to the subsequent time-domain trajectory analysis and micro-Doppler spectrum matching process. This Reni entropy analysis provides rapid pre-screening in the time domain, while the micro-Doppler spectrum provides fine confirmation in the frequency domain, forming a two-stage filtering process that significantly reduces the false alarm rate.

[0043] Furthermore, Raney entropy analysis can be combined with an adaptive threshold algorithm to continuously estimate the statistical characteristics of environmental noise and dynamically update the threshold. For example, the mean and variance of entropy values ​​within a sliding window can be maintained, and the deviation between the current frame entropy value and the historical background model can be calculated. When the deviation exceeds a dynamic threshold, abnormal activity is determined to exist; if the autocorrelation function of the entropy value sequence shows a peak value consistent with a preset oscillation period, it is determined to be periodic human activity. This mechanism can maintain stable recognition performance under different environmental conditions, effectively compensating for the insufficient adaptability of fixed thresholds in complex dynamic scenes.

[0044] Specifically, the process of identifying the user's left-right leg swinging motion based on echo data through point cloud clustering and micro-Doppler spectrum analysis is as follows: First, data preprocessing and target detection are performed: the raw echo data acquired by the radar is processed to generate a range-Doppler matrix containing range and velocity information. Specifically, the millimeter-wave radar transmits a frequency-modulated continuous wave, and by performing a two-dimensional fast Fourier transform on the echo signal, the spectral data in the range and Doppler dimensions are obtained, thereby constructing the range-Doppler matrix. The range dimension resolution depends on the frequency modulation bandwidth.

[0045] After generating the range-Doppler matrix, static clutter filtering is performed. Specifically, moving target display technology is used, which involves differentially processing multiple consecutive frames of the range-Doppler matrix along the slow time dimension to eliminate reflection signals from stationary backgrounds such as the ground, walls, and parked vehicles. Furthermore, a constant false alarm rate (CFAR) detection algorithm can be used to detect targets on the filtered range-Doppler matrix, adaptively setting the detection threshold to extract potential moving target points.

[0046] Preferably, the range of interest (ROI) is set to 0.2 meters to 1 meter, which covers the main range of leg movement when the user is standing behind the vehicle. By processing the distance Doppler data within the ROI as described above, interference from distant targets can be effectively reduced, improving the efficiency and accuracy of subsequent processing.

[0047] Secondly, target classification and leg region localization are performed: the detected point cloud targets are clustered and tracked. The point cloud targets consist of moving target points extracted in the above steps, and each point contains distance, velocity, angle, and echo intensity information. The clustering algorithm can be a density-based spatial clustering algorithm, such as the DBSCAN algorithm, which groups spatially adjacent points with the same velocity into the same target cluster, thereby forming candidate target point cloud clusters.

[0048] Tracking the point cloud clusters obtained by clustering can be achieved using methods such as Kalman filtering or particle filtering. Target associations can be established between consecutive frames to obtain the target's motion trajectory. Tracking can effectively reduce noise and isolated points in single-frame detection and improve the stability of target detection.

[0049] Subsequently, a pre-trained classifier is used to determine whether the target is a pedestrian. The classifier can be a classification model based on machine learning or deep learning. Its input features include the spatial distribution features of point cloud clusters (such as the projection shape of point clouds on horizontal and vertical planes, point cloud density distribution) and micro-Doppler features (such as the bandwidth of the spectrum, energy distribution symmetry, etc.). The classifier is trained on a large number of labeled pedestrian and non-pedestrian samples and can accurately identify pedestrian targets standing behind the rear of the vehicle in complex environments.

[0050] For targets identified as pedestrians, a lower body point cloud cluster is segmented based on point cloud height information. Specifically, a height threshold range is set, such as 0 to 1 meter, and points within this height range are extracted from the pedestrian target's point cloud cluster; these points constitute the lower limb dynamic point cloud cluster. Since the millimeter-wave radar is installed at a height of approximately 0.55 to 0.7 meters, this height range effectively covers the lower legs and lower thighs of standing users, ensuring that the lower limb point cloud cluster has sufficient point density and stability, providing a reliable data foundation for subsequent temporal trajectory analysis. Next, temporal trajectory analysis is performed: the centroid coordinates of the lower limb point cloud cluster in the horizontal direction are calculated in each frame to form a trajectory sequence. The centroid coordinates are calculated by taking the arithmetic mean of the horizontal coordinate values ​​of all points belonging to the lower limb dynamic point cloud cluster in each frame to obtain the horizontal position of the centroid in that frame. The horizontal positions of the centroids in multiple consecutive frames are arranged in chronological order to form a sequence of horizontal displacement trajectories of the centroids.

[0051] The trajectory sequence is bandpass filtered to remove high-frequency noise and low-frequency drift. Specifically, the passband frequency range of the bandpass filter is preferably 0.5Hz to 2.5Hz, which corresponds to the typical frequency of a human leg swinging naturally from side to side. The filter can be implemented as a digital finite impulse response filter or an infinite impulse response filter, and filtering is achieved by convolving or spectral processing of the trajectory sequence in the time or frequency domain.

[0052] Peak and trough detection is performed on the filtered trajectory sequence. A peak is a local maximum value, corresponding to the extreme position of the leg swing to one side (e.g., the right swing limit); a trough is a local minimum value, corresponding to the extreme position of the leg swing to the other side (e.g., the left swing limit). A valid left-right swing cycle consists of a peak and a trough, and the peaks and troughs alternate sequentially. Within a continuous time frame, such as a 2-second time window, if more than a set number of stable cycles are detected (e.g., 3), and each cycle has a reasonable duration (e.g., between 0.4 and 1.5 seconds), and the difference between the peak and trough values ​​of the trajectory (i.e., the swing amplitude) is greater than a preset threshold (e.g., 5 centimeters), then the time-domain trajectory condition is considered satisfied.

[0053] Furthermore, the symmetry of the oscillation amplitude and the consistency of the period can be evaluated. For example, the duration and amplitude differences between adjacent periods can be calculated. If the difference exceeds the preset range, it may be judged as an irregular oscillation that does not meet the time domain conditions.

[0054] Next, frequency domain feature verification is performed: the micro-Doppler spectrum corresponding to the target is extracted synchronously. The micro-Doppler spectrum is obtained by performing a short-time Fourier transform on the target echo data, which reflects the change of Doppler frequency over time caused by micro-motion (such as leg swing) at each scattering point of the target. Specifically, a time segment of one swing cycle is extracted from the original echo data corresponding to the lower limb dynamic point cloud cluster, and time-frequency analysis is performed along the slow time dimension to obtain the micro-Doppler spectrum.

[0055] Within one swing cycle, the signal power or amplitude intensity distribution corresponding to each frequency component in the micro-Doppler spectrum is examined to determine whether it exhibits symmetrical and periodic upper and lower sideband characteristics. When the legs swing left and right, the legs alternately move closer to and further away from the radar, generating positive and negative Doppler frequency shifts respectively. Due to the symmetry of the left-right swing, the spectrum will form approximately symmetrical upper and lower sidebands on both sides of zero frequency. The micro-Doppler spectrum feature extraction and template matching algorithm extracts the corresponding spectral features by detecting the symmetry of the signal power or amplitude intensity distribution corresponding to each frequency component in the micro-Doppler spectrum along the Doppler axis, as well as the periodic changes of the sidebands over time.

[0056] The matching degree of this spectral feature is calculated by comparing it with a preset standard micro-Doppler template for leg swing. The preset template library stores standard micro-Doppler spectral feature templates for users of different heights and swing habits when swinging their legs left and right. The matching degree can be calculated using methods such as normalized cross-correlation, Euclidean distance, and dynamic time warping to calculate the similarity between the real-time extracted spectral features and each template. When the matching degree exceeds a preset matching threshold, it is determined that the frequency domain condition is met.

[0057] This micro-Doppler template matching can effectively eliminate interference from small animal movements and clothing swaying. The micro-Doppler spectrum generated by small animal movements usually appears as a random, irregular narrowband signal, while clothing swaying lacks a symmetrical upper and lower sideband structure. However, the left and right swaying of the legs will form a specific symmetrical pattern on the spectrum. Through template matching, valid user intention actions can be accurately distinguished, significantly reducing the false trigger rate.

[0058] If the above time-domain trajectory analysis and frequency-domain feature verification conditions are met, and the context-aware decision conditions (target distance is less than the preset distance threshold and radial velocity is close to zero) are also met, the user's left and right leg swinging motion is identified, and a command to close the trunk is generated.

[0059] In terms of signal processing, the embodiments of this application also differ fundamentally from existing kicking schemes. Existing kicking schemes primarily focus on simple matching of foot point clouds or trajectories, i.e., detecting whether a foot quickly extends and kicks up from under the bumper. Their judgment criteria are relatively singular and easily affected by accidental disturbances such as small animals darting about, road debris bouncing, or clothing swaying, leading to a high false trigger rate. In contrast, the embodiments of this application employ a signal processing flow that integrates three criteria: temporal trajectory analysis, frequency domain feature verification, and context-aware decision-making. In the temporal domain, by tracking the horizontal displacement trajectory of the lower limb dynamic point cloud cluster's centroid, periodic left-right alternating swing features within the 0.5Hz to 2.5Hz frequency band are extracted and verified multiple times with preset number of cycles, period duration, and swing amplitude. In the frequency domain, by matching the symmetrical sideband features of the micro-Doppler spectrum with a preset template, specific symmetrical spectral patterns formed by leg swings on the spectrogram are effectively eliminated to eliminate interference. In the contextual domain, recognition is only initiated when the user is standing still at a distance of less than 0.5 meters and the radial velocity is close to zero. The combined judgment based on the above three criteria fundamentally changes the signal processing mode of the existing kicking scheme, which relies on a single trigger condition and simple threshold judgment. This significantly reduces the false trigger rate and improves the accuracy and robustness of action recognition.

[0060] This application's embodiments achieve significant technical effects by innovating the action mode of replacing kicking with left and right leg swinging, adapting the radar installation method of replacing the downward detection under the bumper with high-level horizontal detection at the rear of the vehicle, and deeply integrating the signal processing flow of simple trajectory matching with a triple criterion consisting of time-domain trajectory analysis, frequency-domain feature verification, and context-aware decision. These results in more natural user operation, no need to stand on one leg, lower false trigger rate, and stronger environmental adaptability.

[0061] Step 130: In response to the left and right swinging motion of the legs, control the closing of the vehicle's trunk.

[0062] In this embodiment, a trunk closing command is generated based on the left and right swinging motion of the legs, and the trunk closing command is encapsulated into a CAN data frame and sent to the trunk control module. Before executing the trunk closing, the trunk control module performs a safety check to confirm that the trunk is in the open state and the vehicle is in the parking gear or the vehicle speed is lower than a preset safety threshold. During the closing process, anti-pinch detection is performed. When an abnormal increase in resistance is detected, the closing is stopped and the trunk is opened in the reverse direction.

[0063] Specifically, firstly, the execution command is generated and encapsulated using CAN: After the leg swinging motion is successfully verified by the signal processing circuit in the millimeter-wave radar module, the recognition result is sent to the body control module or body domain controller via the internal communication interface. Based on this recognition result, the body control module or body domain controller generates a standard trunk closing control command. This standard trunk closing control command is then encapsulated into a CAN data frame conforming to vehicle communication specifications. The CAN data frame includes standard fields such as a start-of-frame bit, arbitration field, control field, data field, CRC check field, acknowledgment field, and end-of-frame bit. The arbitration field contains an identifier (ID), which determines the transmission priority of the message on the CAN bus. Different functional commands are assigned different identifiers to ensure that the trunk closing command can be correctly recognized and prioritized for transmission on the bus. The data field carries the specific command content, such as the preset encoding value corresponding to the closing command. After encapsulation, the CAN data frame is broadcast to various nodes in the vehicle network via the CAN bus, with the trunk control module acting as the target receiving node and responding to the message.

[0064] Secondly, the execution command reception and safety verification: The trunk control module (usually an independent electronic control unit, i.e., ECU) receives the aforementioned CAN data frames via the CAN bus interface, parses the frames, and extracts the standard trunk closing control command content. Before driving the actuator, the trunk control module performs a final safety check to ensure that all preconditions for executing the closing action are met. The safety check includes the following: confirming that the trunk is currently still open. The trunk control module can determine this by reading the trunk lock status signal or the trunk opening angle sensor signal, ensuring that the trunk status has not changed during command transmission and processing, and avoiding repeated execution of the closing action if the trunk is already closed or an abnormality has occurred. Confirming that the vehicle is currently in a safe operating condition, i.e., the vehicle is in park (P gear) or the vehicle speed is below a preset safety threshold (e.g., 3 km / h), to ensure that the trunk closes when the vehicle is stationary or at low speed, avoiding accidental trunk closing due to misoperation during driving, and confirming that the anti-pinch detection system is in normal working condition and has no fault codes.

[0065] If any of the above safety conditions are not met, the trunk control module will abandon the execution of this closing command and can return abnormal status information via the CAN bus. If all conditions are met, the closing execution phase will begin.

[0066] Then, the closing action and anti-pinch detection are performed: After the safety verification is passed, the trunk control module outputs a drive signal to the electric strut or drive motor through the drive circuit. The electric strut or drive motor is located at the connection between the trunk lid and the vehicle body. The rotation of the motor drives the strut to extend or retract, or the hinge to rotate, thereby driving the trunk lid to rotate downward around the hinge axis to achieve the closing action.

[0067] Throughout the closing process, the anti-pinch detection function operates in real time. Specifically, the trunk control module collects the operating current of the drive motor in real time. When the trunk lid encounters resistance during closing (such as being caught on a user's arm or an object), the motor load increases, and the operating current rises accordingly. If the detected current increase exceeds a preset threshold, it is determined that there is abnormal resistance, which may pose a risk of pinching. Alternatively, a Hall sensor or angle sensor can be installed at the trunk lid hinge to monitor the closing speed and position of the trunk lid in real time. If a sudden drop in speed or stagnation in position is detected during the closing process, it is also determined that there is abnormal resistance.

[0068] If abnormal resistance is detected, the trunk control module immediately stops the closing action and controls the drive motor to run in reverse, causing the trunk lid to open slightly in the opposite direction to release the trapped object and ensure personnel safety. If no abnormal resistance is detected, the closing process continues until the trunk lid is fully closed.

[0069] Finally, the closing process is completed and the system is reset: After the trunk lid is fully closed, the trunk lock body switches from the open state to the locked state. The trunk status detection module detects this change and sends a status signal to the body control module. The trunk control module or the body control module then sends a trunk closing completion status message to the CAN bus to notify other relevant vehicle systems (such as the instrument panel display, anti-theft system, etc.) that the trunk is now closed.

[0070] Subsequently, the millimeter-wave radar module exits the high-power operating mode and enters a low-power standby or hibernation state. The entire vehicle trunk control system resets, clears intermediate data from this interaction, and returns to the initial waiting state, continuing to wait for the next trunk opening event, or entering hibernation to save power based on the overall vehicle operating conditions. This concludes the entire process.

[0071] To adapt to different users' swaying habits and improve the generalization and robustness of recognition, this embodiment employs a lightweight deep learning model for end-to-end classification of time-domain trajectory sequences and micro-Doppler spectral slices. This deep learning model can serve as a replacement or supplement to time-domain trajectory analysis and micro-Doppler template matching, achieving adaptation to different users' swaying habits and further enhancing the generalization and robustness of recognition.

[0072] Furthermore, the specific process of end-to-end classification of time-domain trajectory sequences and micro-Doppler spectral slices using a lightweight deep learning model is as follows: During the training phase, a training sample dataset is first constructed. Radar echo data is collected from users of different heights, body types, clothing styles, and swinging habits when they perform left-right leg swinging movements behind a vehicle. Following the steps described above, the corresponding time-domain trajectory sequences and micro-Doppler spectra are extracted, and these sequences are labeled as valid swinging motion categories. Simultaneously, samples of other non-target movements, such as small animals darting about, pedestrians passing by, clothing swaying, and static standing, are collected and labeled as invalid swinging motion categories. To ensure the model's generalization ability, the training samples should cover environmental factors such as different weather conditions (e.g., sunny days, rainy days), different lighting conditions, and different ground materials. In the data preprocessing stage, the time-domain trajectory sequence and the micro-Doppler spectrum are subjected to length alignment and normalization. For the time-domain trajectory sequence, the centroid horizontal displacement data within a fixed time window (e.g., 2 seconds) can be uniformly extracted, and the time axis is resampled to a fixed length. For the micro-Doppler spectrum, the spectral data within the same time window can be uniformly extracted, and the spectral amplitude is normalized to a preset range. The preprocessed time-domain trajectory sequence and micro-Doppler spectrum slices are used as input to the deep learning model. The lightweight deep learning model adopts a two-branch structure. The first branch is the time-domain feature extraction branch, which is used to process the time-domain trajectory sequence. It can use a one-dimensional convolutional neural network or recurrent neural network structure to extract time-domain features such as periodicity, amplitude, and oscillation rhythm in the trajectory through multiple convolutional layers or recurrent layers. The second branch is the frequency-domain feature extraction branch, which is used to process the micro-Doppler spectrum. It can use a two-dimensional convolutional neural network structure to extract frequency-domain features such as symmetrical sidebands, energy distribution, and periodic changes in the spectrum through multiple convolutional layers. The output features of the two branches are concatenated and then input into a fully connected layer for feature fusion. Finally, the classification result is obtained through the output layer, which is the probability that the input sample belongs to the category of valid swinging motion. During training, the parameters of the deep learning model are updated using the cross-entropy loss function and stochastic gradient descent optimization algorithm. Multiple rounds of iterative training are conducted until the deep learning model achieves the preset accuracy on the validation set. Due to its lightweight network structure, this deep learning model can be deployed in the signal processing circuit of a millimeter-wave radar module or in a vehicle body control module, meeting the real-time performance and computational resource constraints of an automotive embedded environment. During the online inference phase, the temporal trajectory sequence and micro-Doppler spectrum corresponding to the user's actions are collected in real time. After the above preprocessing, these sequences are input into a pre-trained lightweight deep learning model to obtain the probability value of the action being a valid swinging motion. When this probability value exceeds a preset classification threshold, the user's left-right swinging motion of the leg is determined to be recognized.

[0073] This deep learning model, trained end-to-end, can automatically learn deep feature representations of different users' swaying habits without requiring the manual design of complex feature extraction rules, thus improving its adaptability to different user groups. Compared to methods that rely solely on fixed thresholds and template matching, this model has stronger generalization ability and robustness, effectively addressing individual differences in users' swaying amplitude, frequency, and rhythm.

[0074] The vehicle trunk control method provided in this application clusters radar point clouds to separate the dynamic point cloud clusters of the lower limbs, tracks their centroid horizontal displacement trajectory, and analyzes the periodicity of the trajectory, the alternating changes in velocity direction, and the amplitude threshold to determine whether it is a valid swing. This achieves the effect of abandoning the simple proximity triggering method and realizing precise motion capture. By deeply fusing the micro-Doppler spectrum features of the target, the real-time spectrum is matched with a preset standard micro-Doppler template library for leg swing, and combined with the time-domain trajectory for joint judgment, forming a triple criterion of time-domain, frequency-domain, and context. This effectively eliminates interference from small animals running around, clothing swinging, etc., and significantly reduces the false trigger rate. By combining the scene context for comprehensive judgment, subsequent periodic leg movements are only identified as a closing intention when the target distance is less than 0.5 meters and the relative radial velocity is close to zero, i.e., when the user is relatively stationary. This achieves the effect of avoiding false triggering when pedestrians pass by. By introducing a lightweight deep learning model, end-to-end classification of time-domain trajectory sequences and micro-Doppler spectrum slices is performed to adapt to different user swing habits and further improve the recognition generalization ability and robustness.

[0075] This application also provides a vehicle trunk control system, see below. Figure 2 As shown, the vehicle trunk control system provided in this application embodiment includes: The status detection module 210 is used to activate the millimeter-wave radar located at the rear of the vehicle in response to the opening of the vehicle's trunk. The data processing module 220 is used to identify the user's left and right leg swinging motion based on the echo data collected by millimeter-wave radar through point cloud clustering and micro-Doppler spectrum analysis. The left and right leg swinging motion is the reciprocating swinging of the user's legs in the horizontal direction. The control execution module 230 is used to control the closing of the vehicle's trunk in response to the left and right swinging motion of the legs.

[0076] It should be noted that the principle of the vehicle trunk control system provided in this application embodiment to solve the technical problem is similar to that of the vehicle trunk control method provided in this application embodiment. Therefore, the implementation of the vehicle trunk control system provided in this application embodiment can refer to the implementation of the vehicle trunk control method provided in this application embodiment, and the repeated parts will not be described again.

[0077] After introducing the vehicle trunk control method and device provided in the embodiments of this application, the electronic device provided in the embodiments of this application will be briefly introduced next.

[0078] See Figure 3 As shown, the electronic device 500 provided in this application embodiment includes at least a processor 501, a memory 502, and a computer program stored in the memory 502 and executable on the processor 501. When the processor 501 executes the computer program, it implements the vehicle trunk control method provided in this application embodiment.

[0079] The electronic device 500 provided in this application embodiment may further include a bus 503 connecting different components (including processor 501 and memory 502). The bus 503 represents one or more types of bus structures, including memory bus, peripheral bus, local area bus, etc.

[0080] Memory 502 may include a readable storage medium in the form of volatile memory, such as random access memory (RAM) 5021 and / or cache memory 5022, and may further include read-only memory (ROM) 5023. Memory 502 may also include a program tool 5025 having a set (at least one) of program modules 5024, including but not limited to an operating subsystem, one or more application programs, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.

[0081] Processor 501 can be a single processing element or a collective term for multiple processing elements. For example, processor 501 can be a central processing unit (CPU) or one or more integrated circuits configured to implement the vehicle trunk control method provided in the embodiments of this application. Specifically, processor 501 can be a general-purpose processor, including but not limited to CPUs, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0082] Electronic device 500 can communicate with one or more external devices 504 (e.g., keyboard, remote control, etc.), and also with one or more devices that enable a user to interact with electronic device 500 (e.g., mobile phone, computer, etc.), and / or with devices that enable electronic device 500 to communicate with one or more other electronic devices 500 (e.g., router, modem, etc.). This communication can be performed through input / output (I / O) interface 505. Furthermore, electronic device 500 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) through network adapter 506. Figure 3 As shown, network adapter 506 communicates with other modules of electronic device 500 via bus 503. It should be understood that, although... Figure 3 As not shown, other hardware and / or software modules may be used in conjunction with the electronic device 500, including but not limited to microcode, device drivers, redundant processors, external disk drive arrays, Redundant Arrays of Independent Disks (RAID) subsystems, tape drives, and data backup storage subsystems.

[0083] It should be noted that, Figure 3 The electronic device 500 shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0084] The following describes the computer-readable storage medium provided in the embodiments of this application. The computer-readable storage medium provided in the embodiments of this application stores computer instructions, which, when executed by a processor, implement the vehicle trunk control method provided in the embodiments of this application. Specifically, the computer instructions can be built into or installed in the processor, so that the processor can implement the vehicle trunk control method provided in the embodiments of this application by executing the built-in or installed computer instructions.

[0085] In addition, the vehicle trunk control method provided in this application embodiment can also be implemented as a computer program product, which includes program code. The program code implements the vehicle trunk control method provided in this application embodiment when it is run on a processor.

[0086] The computer program product provided in this application embodiment may employ one or more computer-readable storage media, which may be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination thereof. Specifically, more specific examples (a non-exhaustive list) of computer-readable storage media include electrical connections with one or more wires, portable disks, hard disks, RAM, ROM, erasable programmable read-only memory (EPROM), optical fibers, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0087] The computer program product provided in this application embodiment can be a CD-ROM and include program code, and can also run on electronic devices such as computers. However, the computer program product provided in this application embodiment is not limited thereto. In this application embodiment, the computer-readable storage medium can be any tangible medium that contains or stores program code, which can be used by or in conjunction with an instruction execution system, device, or apparatus.

[0088] It should be noted that although several units or sub-units of the device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of this application, the features and functions of two or more units described above can be embodied in one unit. Conversely, the features and functions of one unit described above can be further divided and embodied by multiple units.

[0089] Furthermore, although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0090] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0091] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of the embodiments of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of the claims of this application and their equivalents, this application also intends to include these modifications and variations.

Claims

1. A method for controlling a vehicle's trunk, characterized in that, include: The millimeter-wave radar located at the rear of the vehicle is activated in response to the opening of the vehicle's trunk. Based on the echo data collected by the millimeter-wave radar, the user's left and right leg swinging motion is identified by point cloud clustering and micro-Doppler spectrum analysis, wherein the left and right leg swinging motion is the reciprocating swinging of the user's leg in the horizontal direction. In response to the left and right swinging motion of the legs, the vehicle's trunk is closed.

2. The vehicle trunk control method according to claim 1, characterized in that, The echo data is point cloud data; based on the echo data collected by the millimeter-wave radar, the user's left-right leg swinging motion is identified through point cloud clustering and micro-Doppler spectrum analysis, including: The point cloud data is clustered to obtain point cloud clusters; Based on the vertical coordinate information in the point cloud data, the lower limb dynamic point cloud cluster is separated from the point cloud cluster; Determine the horizontal displacement trajectory of the centroid of the lower limb dynamic point cloud cluster, and obtain the user's micro-Doppler spectrum based on the echo data corresponding to the lower limb dynamic point cloud cluster; When the horizontal displacement trajectory of the centroid has a preset number of alternating left and right swing cycles within a preset time period and the swing amplitude is greater than a preset amplitude threshold, and the matching degree between the micro-Doppler spectrum and the preset leg swing template exceeds a preset matching threshold, the user's left and right leg swinging motion is determined to be identified.

3. The vehicle trunk control method according to claim 1 or 2, characterized in that, Before identifying the user's left-right leg swinging motion based on the echo data collected by the millimeter-wave radar through point cloud clustering and micro-Doppler spectrum analysis, the process also includes: Based on the distance information in the echo data, the distance between the user and the vehicle is determined; When the distance is less than a preset distance threshold and the absolute value of the user's radial velocity relative to the vehicle is less than a preset speed threshold, the user's leg swinging motion is identified.

4. The vehicle trunk control method according to claim 2, characterized in that, Determining the horizontal displacement trajectory of the centroid of the lower limb dynamic point cloud cluster includes: Determine the centroid coordinates of the lower limb dynamic point cloud cluster in the horizontal direction in each frame to form a trajectory sequence; The trajectory sequence is subjected to bandpass filtering, and peak and valley values ​​are detected in the filtered trajectory sequence; wherein, an effective left-right swinging motion cycle of the leg includes a local maximum value and a local minimum value of the horizontal displacement trajectory.

5. The vehicle trunk control method according to claim 2, characterized in that, The matching degree between the micro-Doppler spectrum and the preset leg swing template exceeds a preset matching threshold, including: Extract the spectral features of the micro-Doppler spectrum and check whether the spectral features exhibit symmetrical and periodic upper and lower sideband features; The matching degree between the spectral features and the standard spectral features in the preset leg swing template is calculated. When the matching degree exceeds the preset matching threshold, it is determined that the frequency domain condition is met.

6. The vehicle trunk control method according to claim 1, characterized in that, In response to the left-right swinging motion of the leg, controlling the closing of the vehicle's trunk includes: Based on the left and right swinging motion of the legs, a command to close the trunk is generated; The command to close the trunk is encapsulated into a CAN data frame and sent to the trunk control module. Before closing the trunk, the trunk control module performs a safety check to confirm that the trunk is open and the vehicle is in park or the vehicle speed is below a preset safety threshold. During the closing process, an anti-pinch detection is performed. If an abnormal increase in resistance is detected, the closing process stops and the device is opened in reverse.

7. The vehicle trunk control method according to claim 1, characterized in that, After activating the millimeter-wave radar located at the rear of the vehicle, the following is also included: The indicator lights located inside the trunk or at the rear of the vehicle are put into a guidance state to guide the user to the interaction position directly behind the rear of the vehicle.

8. A vehicle trunk control system, characterized in that, include: The status detection module is used to activate the millimeter-wave radar located at the rear of the vehicle in response to the opening of the vehicle's trunk. The data processing module is used to identify the user's left and right leg swinging motion based on the echo data collected by the millimeter-wave radar through point cloud clustering and micro-Doppler spectrum analysis, wherein the left and right leg swinging motion is the reciprocating swinging of the user's leg in the horizontal direction. The control execution module is used to control the closing of the vehicle trunk in response to the left and right swinging motion of the legs.

9. An electronic device, characterized in that, The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the vehicle trunk control method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the vehicle trunk control method as described in any one of claims 1 to 7.

Images