Vehicle tail door opening method, device and system and vehicle

Abstract

The invention relates to the technical field of vehicles, and discloses a tail door opening method, device and system of a vehicle and the vehicle. The method comprises the steps that an image sequence of the feet of a user is collected; based on the current frame of the image sequence, determining space coordinates and motion feature data of the moving foot and the standing foot of the user at the current moment; according to the space coordinates and the motion feature data, judging whether to trigger the transfer from the current state to the next state or not through a kicking action state machine model; when the state of the kicking action state machine is transferred to the judgment state, a set of space key points recorded in the state transferring process are obtained; and if the spatial geometrical relationship among the group of spatial key points meets a preset constraint condition, it is judged that an effective kicking action occurs, and a tail door of the vehicle is controlled to be opened. According to the technical scheme, the problems of low recognition rate and low reliability of tail gate opening triggering in related technologies are solved, and the accuracy and reliability of tail gate opening triggering are improved.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, specifically to a method, device, system, and vehicle for opening the tailgate. Background Technology

[0002] With the development of automotive intelligence, users' demand for convenient tailgate opening functions is also increasing. Among them, the foot-operated sensor tailgate opening technology (commonly known as "one-touch opening") is highly favored due to its unique convenience when holding items with both hands.

[0003] In related technologies, capacitive sensors or microwave radar sensors are used to detect when a user's foot approaches a specific trigger area at the rear of the vehicle to trigger the tailgate opening command. In real-world usage environments, false triggers can easily occur due to factors such as ambient humidity, the proximity of metal objects, or the passage of small animals, affecting user experience and even posing safety hazards. Furthermore, tailgate opening and closing schemes based on deep learning algorithms to recognize foot movements are prone to false triggers, making it difficult to guarantee their reliability.

[0004] In other words, the tailgate opening triggering technology suffers from low recognition rate and low reliability. Summary of the Invention

[0005] This application provides a method, device, system, and vehicle for opening the tailgate of a vehicle, in order to solve the problems of low recognition rate and low reliability in the triggering of tailgate opening in related technologies.

[0006] In a first aspect, this application provides a method for opening the tailgate of a vehicle, the method comprising:

[0007] The system uses an in-vehicle vision sensor to capture image sequences of the user's feet. Based on the current frame of the image sequence, determine the spatial coordinates and motion feature data of the user's moving foot and standing foot at the current moment; Based on spatial coordinates and motion characteristic data, the kicking motion state machine model determines whether to trigger the transition from the current state to the next state. The kicking motion state machine model predefines multiple node states, which include at least the initial state, ready state, raised state, sweeping state, falling state, and judgment state. When the state machine of the kicking action transitions to the decision state, a set of spatial key points recorded during the state transition is obtained. The set of spatial key points includes at least the starting point of the kicking foot in the ready state, the highest point of the kicking foot in the lifting state, and the farthest point of the kicking foot in the sweeping state. If the spatial geometric relationship between a set of key points satisfies the preset constraints, a valid kicking action is determined to have occurred, and the tailgate of the vehicle is opened.

[0008] In one optional implementation, based on spatial coordinates and motion characteristic data, a kicking motion state machine model is used to determine whether a transition from the current state to the next state is triggered, including: In the initial state, when the position of the moving foot is detected to be within the preset effective area and remains stationary, the system transitions to the ready state. In the ready state, when a change in the position of the moving foot is detected, the current position is recorded as the starting point of the movement, and the movement is transferred to the raised state; In the raised state, when the vertical displacement of the moving foot meets the first condition, the highest point of the raised position is recorded or updated, and the process is transferred to the sweeping state. In the sweeping state, when the horizontal displacement of the moving foot meets the second condition, the furthest point of the sweep is recorded or updated, and the process is transferred to the fallback state. In the falling state, when the moving foot is detected to have fallen within the preset height range, the process transitions to the judgment state.

[0009] In one optional implementation, when the vertical displacement of the moving foot satisfies a first condition, the highest point of lifting is recorded or updated, including: Obtain the reference space coordinates of the starting point of the action; Calculate the real-time vertical displacement of the moving foot relative to the starting point of the movement based on the spatial coordinates and the reference spatial coordinates. If the real-time vertical displacement is greater than the first threshold at the first moment, and the vertical movement speed of the moving foot in the motion feature data is less than the first speed threshold, it is determined that the vertical displacement of the spatial coordinates meets the first condition, and the three-dimensional coordinates of the spatial coordinates at the first moment are recorded as the highest point of lifting.

[0010] In one optional implementation, when a horizontal displacement of spatial coordinates is detected that satisfies a second condition, the furthest point of the sweep is recorded or updated, including: Obtain the reference space coordinates of the starting point of the action; Calculate the real-time horizontal displacement of the moving foot relative to the starting point of the movement based on the spatial coordinates and the reference spatial coordinates. If the real-time horizontal displacement is greater than the second threshold at the second moment, and the horizontal movement speed of the moving foot in the motion feature data is less than the second velocity threshold, it is determined that the displacement of the spatial coordinates in the horizontal direction satisfies the second condition, and the three-dimensional coordinates of the spatial coordinates at the second moment are recorded as the farthest point of the sweep.

[0011] In one optional implementation, if the spatial geometric relationship between a set of spatial key points satisfies preset constraints, a valid kicking action is determined to have occurred, including: Calculate the vertical displacement of the highest point of the lift relative to the starting point of the movement; Calculate the horizontal displacement of the furthest point of the sweep relative to the starting point of the action; If the vertical and horizontal displacements meet the preset constraints, a valid kicking action is determined to have occurred. The preset constraints include a vertical displacement greater than zero and a horizontal displacement pointing towards the rear of the vehicle.

[0012] In one alternative implementation, the method further includes: When the kicking motion state machine model is in any of the states of lifting, sweeping, and falling, if any of the following situations are detected based on the current frame of the image sequence, the kicking motion state machine model will be reset to the initial state: The spatial coordinates of the moving foot were detected to be outside the preset effective detection area. The duration of the current state exceeds the maximum time threshold preset for the current state; Displacement of the standing foot exceeding the stability threshold was detected.

[0013] In one alternative implementation, after determining that a valid kick has occurred, the method further includes: Reset the current state of the kicking action state machine model to the initial state.

[0014] Secondly, this application provides a tailgate opening device for a vehicle, the device comprising: The acquisition module is used to acquire image sequences of the user's feet through an onboard vision sensor; The determination module is used to determine the spatial coordinates and motion feature data of the user's feet at the current moment based on the current frame of the image sequence. The user's feet include the moving foot and the standing foot. The judgment module is used to determine whether to trigger the transition from the current state to the next state based on the spatial coordinates and motion feature data through the kicking motion state machine model. The kicking motion state machine model predefines multiple node states, which include at least the initial state, the ready state, the raised state, the sweeping state, the falling state, and the judgment state. The acquisition module is used to acquire a set of spatial key points recorded during the state transition when the state machine of the kicking action transitions to the judgment state. The set of spatial key points includes at least the starting point of the movement of the foot in the ready state, the highest point of the foot in the lifting state, and the farthest point of the sweeping state. The tailgate opening module is used to determine that a valid kicking action has occurred and to control the opening of the vehicle's tailgate if the spatial geometric relationship between a set of key spatial points meets preset constraints.

[0015] Thirdly, this application provides a vehicle tailgate opening system, including: an on-board image acquisition device, an on-board data bus, an image processing and decision unit, a body control unit, and a tailgate actuator. The on-board data bus connects the on-board image acquisition device and the image processing and decision unit, and also connects the image processing and decision unit, the body control unit, and the tailgate actuator. The on-board image acquisition device is used to acquire image sequences of a user's feet. The body control unit controls the tailgate actuator to open the tailgate. The image processing and decision unit is configured to perform the vehicle tailgate opening method of the first aspect or any corresponding embodiment described above.

[0016] Fourthly, this application provides a computer-readable storage medium storing computer instructions for causing a computer to execute the tailgate opening method of a vehicle according to the first aspect or any corresponding embodiment described above.

[0017] According to the tailgate opening method of a vehicle provided in this application, the following beneficial technical effects can be achieved compared with the prior art: By acquiring image sequences of the user's feet using an in-vehicle vision sensor, continuous monitoring of foot movement is achieved, while leveraging existing in-vehicle hardware to reduce system cost and complexity. Based on the current frame of the image sequence, the spatial coordinates and motion characteristics of the user's moving and standing feet at the current moment are determined, providing accurate and state-discriminative real-time input data for the state machine. By distinguishing between the moving and standing feet, the accuracy of action intent judgment is improved from the data source. Based on the spatial coordinates and motion characteristics, a kicking action state machine model determines whether to trigger a transition from the current state to the next state. This kicking action state machine model predefines multiple node states, including at least an initial state, a ready state, a raised state, a sweeping state, a falling state, and a judgment state. By decomposing continuous actions into discrete state sequences with clear physical meanings and performing rigorous temporal logic judgments, the system achieves this. This system enables refined real-time tracking and preliminary filtering of the action process, effectively eliminating incomplete or out-of-sequence interference actions. When the kicking action state machine transitions to the judgment state, it acquires a set of spatial key points recorded during the state transition. This set of spatial key points includes at least the highest point of the raised foot in the raised state and the farthest point of the swept foot in the swept state. It extracts key trajectory spatial feature data dynamically recorded during the action, providing a physically meaningful basis for final verification. If the spatial geometric relationship between the set of spatial key points satisfies preset constraints, a valid kicking action is determined, and the vehicle's tailgate is opened. By introducing secondary verification based on spatial geometric rules, a final and rigorous physical rationality check is constructed, ensuring that only actions that simultaneously meet the correct timing and correct spatial form can be triggered, thus achieving an extremely low false trigger rate. This improves the accuracy and reliability of tailgate opening triggering. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 This is a flowchart of a method for opening the tailgate of a vehicle according to an embodiment of this application; Figure 2 This is a flowchart of another method for opening the tailgate of a vehicle according to an embodiment of this application; Figure 3 This is a schematic diagram of a kicking motion state machine model according to an embodiment of this application; Figure 4This is a schematic diagram of a kicking motion trajectory according to an embodiment of this application; Figure 5 This is a schematic diagram of a kicking action state machine model for recognizing kicking actions according to an embodiment of this application; Figure 6 This is a schematic diagram of the state transition of a kicking action state machine model according to an embodiment of this application; Figure 7 This is a structural block diagram of a vehicle tailgate opening device according to an embodiment of this application; Figure 8 This is a schematic diagram of the configuration of a vehicle tailgate opening system according to an embodiment of this application; Figure 9 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] It is understood that before using the technical solutions disclosed in the various embodiments of this application, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this application in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0022] 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. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "a plurality of" means two or more, unless otherwise explicitly specified.

[0023] With the deep integration of the automotive industry and electronic information technology, automotive intelligence has become an irreversible development trend. Within this trend, in-vehicle vision systems, especially rearview cameras, have gradually moved from being high-end optional features to becoming standard equipment in most new cars.

[0024] Meanwhile, user demand for convenient vehicle opening functions is also growing. Among these, foot-activated tailgate opening technology (commonly known as "one-touch opening") has emerged. In related technologies, foot-activated solutions mostly rely on capacitive sensors or microwave radar sensors. Capacitive sensors trigger opening and closing commands by detecting changes in capacitance caused by foot proximity. However, this technology has significant drawbacks in practical applications: First, its anti-interference capability is poor, and it is prone to false triggers due to environmental humidity, the proximity of metal objects, or the passage of small animals, affecting user experience and even posing safety hazards; second, the detection area and sensitivity adjustment of capacitive sensors are not flexible enough, resulting in insufficient accuracy in recognizing foot movements; furthermore, the hardware design and installation debugging required to achieve stable detection also increase the overall system cost. On the other hand, tailgate opening and closing solutions based on deep learning algorithms to recognize foot movements are prone to false triggers, making it difficult to guarantee their reliability.

[0025] According to an embodiment of this application, a method for opening the tailgate of a vehicle is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0026] This embodiment provides a method for opening the tailgate of a vehicle, which can be used in the vehicle's central domain controller or on-board host. Figure 1 This is a flowchart of a method for opening the tailgate of a vehicle according to an embodiment of this application, as shown below. Figure 1 As shown, the process includes the following steps: Step S101: Acquire image sequences of the user's feet using the vehicle-mounted vision sensor.

[0027] Specifically, an automotive vision sensor refers to an image acquisition device installed at the rear of a vehicle to capture images of the environment behind the vehicle. It can be a color camera, a wide-angle camera or depth camera specifically added for foot-activated sensing, and preferably an existing reversing camera. Its main purpose is to continuously acquire images of the user's foot area at a certain frame rate (e.g., 10Hz or higher). An image sequence refers to multiple frames (i.e., a video stream) continuously acquired and output by the automotive vision sensor in chronological order. This sequence constitutes the raw visual data source required for subsequent processing, reflecting the continuous movement of the foot over a period of time.

[0028] The system utilizes in-vehicle cameras to capture real-time video streams of user foot movements. It directly reuses or integrates with existing vehicle visual perception hardware at low cost, eliminating the need for additional dedicated sensors (such as radar). This significantly reduces system hardware costs and layout complexity, and provides a data foundation for subsequent real-time motion recognition based on image sequences.

[0029] Step S102: Based on the current frame of the image sequence, determine the spatial coordinates and motion feature data of the user's moving foot and standing foot at the current moment.

[0030] Specifically, the current frame refers to the most recently acquired and yet-to-be-processed image in the image sequence, representing the visual information at the "current moment." Spatial coordinates refer to the three-dimensional coordinates (X, Y, Z) of the user's feet (especially key parts such as the ankle or toes) in the vehicle coordinate system, for example, obtained through monocular visual geometric estimation or direct acquisition by a depth camera. Motion feature data refers to parameters describing the foot's motion state calculated based on the current frame and historical frame data, such as movement speed, direction of movement, and displacement relative to the previous frame. A moving foot refers to a foot whose position changes significantly in a continuous image sequence, i.e., the foot performing a kicking motion. A standing foot refers to the other foot that remains relatively still or essentially immobile in a continuous image sequence, supporting the user's body.

[0031] The system processes the latest frame of image in real time, first identifying both feet and dynamically distinguishing them into "moving feet" and "standing feet" based on their motion state. The moving foot is the performer of the action, while the standing foot serves as a stability reference for the moving foot. The precise three-dimensional position and dynamic parameters of the moving foot, such as velocity and displacement, are calculated in real time, providing accurate and state-discriminative input data for the subsequent state machine. Distinguishing between both feet allows the system to incorporate the stability of the standing foot as a crucial basis for determining the action intent, improving the accuracy and anti-interference capability of action recognition from the data source. For example, it effectively eliminates misjudgments caused by alternating foot movements such as "walking."

[0032] Preferably, based on the difference in motion vectors between the two feet in the image sequence, the foot with the larger displacement is identified as the moving foot, and the foot with the smaller displacement is identified as the standing foot.

[0033] It should be noted that the moving foot and standing foot are not tied to the user's left and right feet, but are updated in real time based on the image recognition results.

[0034] Step S103: Based on the spatial coordinates and motion feature data, determine whether to trigger the transition from the current state to the next state through the kicking motion state machine model. The kicking motion state machine model predefines multiple node states, which include at least the initial state, preparation state, lifting state, sweeping state, falling state, and judgment state.

[0035] Specifically, the kicking motion state machine model refers to a finite state machine used to identify kicking motions. It consists of predefined states, state transition conditions, and actions, and is used to simulate and identify the temporal stages of a complete kicking motion. Based on real-time input spatial coordinates and motion characteristic data, it determines whether the conditions are met in the current "state" and switches to the next "state." The initial state refers to the starting state after the system is powered on or reset, where it waits for the target to enter. The ready state refers to the state where the kicking foot is detected entering the effective area and remains stationary, indicating that the action is about to begin. The lifting state refers to the state where the kicking foot begins to move vertically upwards. The sweeping state refers to the state where the kicking foot completes its upward lift and begins to move horizontally towards the rear of the vehicle. The falling state refers to the state where the kicking foot completes its forward kick, begins to move downwards, and retracts. The decision state is the final decision state entered after the action sequence is completed, where the recorded key trajectory points are finally verified spatially and geometrically.

[0036] This step drives the state machine's flow in real time based on the current spatial coordinates and motion data. The system continuously determines whether the latest data meets the conditions for transitioning from the "current state" to the "next state," such as displacement and velocity thresholds. By decomposing continuous and complex foot movements into discrete state sequences with clear physical meanings and establishing strict transition conditions, the system achieves refined and interpretable real-time tracking and filtering of the action process. This lays the logical foundation for accurately determining valid actions and effectively eliminates a large number of incomplete or out-of-sequence interference actions.

[0037] Step S104: When the state of the kicking action state machine transitions to the determination state, a set of spatial key points recorded during the state transition is obtained. The set of spatial key points includes at least the highest point of the kicking foot when it is in the lifting state and the farthest point of the kicking foot when it is in the sweeping state.

[0038] Specifically, spatial key points refer to the spatial coordinates that are marked and recorded in real time during state transitions, representing the specific trajectory characteristics of the moving foot. For example, in addition to the highest and farthest points, they usually include the starting point of the action recorded in the preparation state. The highest point of lifting refers to the three-dimensional coordinates corresponding to the maximum height reached by the moving foot in the vertical direction (Y-axis, perpendicular to the ground and upwards) during the lifting state, through continuous comparison and updating. The farthest point of sweeping refers to the three-dimensional coordinates corresponding to the position of the maximum displacement of the moving foot in the horizontal direction (Z-axis, from the rear of the vehicle to the front of the vehicle) during the sweeping state, through continuous comparison and updating.

[0039] After the state transition of the kicking motion state machine is completed, the trajectory feature points dynamically recorded in each previous state stage are summarized and extracted. It should be noted that these points are captured and updated in real time during the action, rather than being extracted afterward from a fitted complete trajectory curve. This provides physically meaningful trajectory feature data for the final verification step. These key points collectively define the core contour of the kicking motion in space, enabling efficient and accurate evaluation of the motion's amplitude, direction, and rationality without the need to process large amounts of raw data.

[0040] Step S105: If the spatial geometric relationship between a set of spatial key points satisfies the preset constraint conditions, then it is determined that a valid kicking action has occurred, and the tailgate of the vehicle is opened.

[0041] Specifically, preset constraints refer to a series of pre-defined geometric rules used to determine whether the trajectory of a foot movement is spatially reasonable. These rules are based on the physical characteristics of a valid kick. For example, vertical constraints: the vertical position of the highest point of the kick must be higher than the starting point of the movement (indicating that the foot was indeed lifted upwards). Horizontal constraints: the horizontal coordinate of the furthest point of the sweep in the vehicle coordinate system must be located in the direction from the rear of the vehicle to the front of the vehicle (indicating that the foot kicked towards the rear of the vehicle, not laterally or in any other direction). Optionally, amplitude constraints may also be included: the vertical distance between the highest point of the kick and the starting point of the movement, and the horizontal distance between the furthest point of the sweep and the starting point of the movement, must each be greater than a minimum threshold (ensuring sufficient amplitude of the movement and excluding slight wobbling).

[0042] This step rigorously verifies the physical plausibility of the "action contour," a set of spatial key points, identified by the kicking motion state machine. The system substitutes the extracted key points into preset geometric rules for calculation and judgment. A second, precise filter based on spatial logic is introduced on top of the state transition conditions. Even if an action passes the temporal filter of the state machine, it will be rejected at this stage if its spatial form does not conform to the biomechanical laws of human kicking (such as simply raising the leg high in place or swinging it sideways). This greatly reduces the false trigger rate and ultimately achieves the high-reliability control goal of "only actions that simultaneously satisfy the correct temporal sequence and correct spatial form will trigger the tailgate opening."

[0043] The tailgate opening method for a vehicle provided in this embodiment acquires image sequences of the user's feet using an onboard vision sensor, achieving continuous monitoring of the foot movement process and reducing system cost and complexity by utilizing existing onboard hardware. Based on the current frame of the image sequence, the spatial coordinates and motion feature data of the user's moving foot and standing foot at the current moment are determined, providing accurate and state-discriminative real-time input data for the state machine. By distinguishing between the moving foot and the standing foot, the accuracy of action intent judgment is improved from the data source. Based on the spatial coordinates and motion feature data, a kicking action state machine model is used to determine whether to trigger a transition from the current state to the next state. The kicking action state machine model predefines multiple node states, including at least an initial state, a ready state, a raised state, a sweeping state, a falling state, and a judgment state. By decomposing the continuous action into discrete state sequences with clear physical meaning and performing... Strict timing logic judgment enables refined real-time tracking and preliminary filtering of the action process, effectively eliminating incomplete or timing-inappropriate interference actions. When the kicking action state machine transitions to the judgment state, a set of spatial key points recorded during the state transition is acquired. This set of spatial key points includes at least the highest point of the raised foot in the raised state and the farthest point of the swept foot in the swept state. This extracts key trajectory spatial feature data dynamically recorded during the action, providing a physically meaningful basis for final verification. If the spatial geometric relationship between the set of spatial key points satisfies preset constraints, a valid kicking action is determined, and the vehicle's tailgate is opened. By introducing secondary verification based on spatial geometric rules, a final and rigorous physical rationality check is constructed, ensuring that only actions that simultaneously satisfy the correct timing and correct spatial form can be triggered, thus achieving an extremely low false trigger rate. This improves the accuracy and reliability of tailgate opening triggering.

[0044] This embodiment provides a method for opening the tailgate of a vehicle, which can be used in the central domain controller or on-board host of the vehicle described above. Figure 2 This is a flowchart of another method for opening the tailgate of a vehicle according to an embodiment of this application, such as... Figure 2 As shown, the process includes the following steps: Step S201: Acquire image sequences of the user's feet using the vehicle-mounted vision sensor.

[0045] Please see details Figure 1 Step S101 of the illustrated embodiment will not be described again here.

[0046] Step S202: Based on the current frame of the image sequence, determine the spatial coordinates and motion feature data of the user's moving foot and standing foot at the current moment.

[0047] Please see details Figure 1Step S102 of the illustrated embodiment will not be described again here.

[0048] Step S203: Based on the spatial coordinates and motion feature data, determine whether to trigger the transition from the current state to the next state through the kicking motion state machine model. The kicking motion state machine model predefines multiple node states, which include at least the initial state, the ready state, the raised state, the sweeping state, the falling state, and the judgment state.

[0049] Specifically, step S203 above includes the following steps: Step S2031: In the initial state, when the position of the moving foot is detected to be within the preset effective area and remains stationary, the process transitions to the preparation state.

[0050] Specifically, the preset effective area refers to a predefined three-dimensional spatial range behind the vehicle. For example, it could be located near the bottom of the rear bumper, such as a cuboid or ellipsoidal spatial area. This area is used to define the initial trigger range for kicking motion recognition.

[0051] The system only begins to continuously track and assess the status of a user's foot when the user's foot enters this area, thereby avoiding responding to irrelevant actions that are far from the vehicle. This effectively narrows the detection range, reduces the possibility of false triggers, and lowers the system's computational load.

[0052] Step S2032: In the ready state, when a change in the position of the moving foot is detected, the current position is recorded as the starting point of the action, and the movement is transferred to the raised state.

[0053] Specifically, the starting point of an action refers to the spatial coordinates (usually three-dimensional coordinates) of the moving foot recorded at the moment the system first detects that the moving foot has left the stationary state and begun to move in the ready state. It marks the spatial starting reference of an action to be identified.

[0054] The system continuously monitors in the ready state, and once the moving foot begins to move, its current position is immediately recorded as the "action start point." It captures the initial moment of the action and establishes a spatial reference baseline. This provides an accurate reference origin for all subsequent spatial displacement calculations (such as vertical lifting height and horizontal sweeping distance), enabling the quantification of the action amplitude. Combined with the condition of "entering the effective area and remaining stationary," it ensures that the system only tracks conscious initiation actions starting from a stationary state, effectively filtering out those already moving or swaying within the area, thus improving the accuracy of intent recognition.

[0055] Step S2033: In the raised state, when the vertical displacement of the moving foot meets the first condition, record or update the highest point of the raised position and switch to the sweeping state.

[0056] Specifically, the first condition refers to the preset quantitative rules used to determine the completion of the "lifting" action and allow entry into the next stage. These typically include displacement conditions and motion trend conditions. The displacement condition is that the real-time vertical displacement of the moving foot relative to the starting point of the action reaches or exceeds a preset displacement threshold. This threshold ensures that the action has a clear "lifting upward" intention and sufficient amplitude to filter out minor leg swaying. The motion trend condition is that, based on the vertical displacement meeting the threshold, motion characteristic data (such as velocity) shows that the movement of the moving foot in the vertical direction has slowed down, stopped, or turned downward. For example, the vertical velocity is less than a small velocity threshold or changes from positive to negative. This indicates that the foot has reached the peak of vertical movement or has begun to change direction. The spatial coordinates at this point are recorded, or the previously recorded three-dimensional coordinates of the highest point of the lift are updated.

[0057] This step determines the completion of the lifting phase by monitoring the height of the raised foot to meet the required amplitude, allowing the state machine to transition from the current lifting state to the next sweeping state. This achieves precise capture and phased judgment of the "lifting" action, ensuring the strictness of state transitions and preventing premature entry into subsequent stages due to accidental or incomplete lifting movements. Dynamically recording the highest point of the lift provides crucial data points for subsequent spatial geometric relationship determination, a key element in ensuring the system can accurately distinguish between "intentional kicking actions" and "unintentional leg lifting or stepping," thus improving the overall reliability of the recognition.

[0058] Step S2034: In the sweep state, when the horizontal displacement of the moving foot is detected to meet the second condition, record or update the furthest point of the sweep and transfer to the fall state.

[0059] Specifically, the second condition refers to the preset quantitative rules used to determine the completion of the "sweeping" motion and allow entry into the next stage. It typically includes two criteria: displacement condition and motion trend condition. The displacement condition refers to the real-time horizontal displacement of the foot relative to the starting point of the motion reaching or exceeding a preset displacement threshold. This threshold ensures the motion has a clear intention to "kick towards the rear of the car" and sufficient amplitude. The motion trend condition refers to the situation where, based on the horizontal displacement meeting the threshold, motion characteristic data (such as speed) shows that the foot's horizontal movement (from the rear of the car to the front) has slowed down, stopped, or reversed (e.g., the horizontal speed is less than a small speed threshold or changes from positive to negative). This indicates that the foot has reached its furthest point in the horizontal direction or has begun to retract.

[0060] During the sweeping phase, the system continuously monitors and updates the furthest point of the sweep. Only when the horizontal movement simultaneously meets the amplitude requirements and the kinematic characteristics of "reaching the furthest point" is the sweeping phase considered complete, allowing the state machine to transition to the next phase (the retracement phase). This ensures the strictness of the state transition and avoids premature entry into the retracement phase due to an incomplete backward kick. Dynamically recording the furthest point of the sweep provides another crucial data point for subsequent spatial geometric relationship determination. The second condition, in conjunction with the first, constitutes a precise decomposition and verification of the complete "kick" action (first lifting, then kicking forward), and is one of the core mechanisms enabling the system to reliably identify valid kicking actions and eliminate interference from other directional movements.

[0061] Step S2035: In the falling state, when the moving foot is detected to fall within the preset height range, the process transitions to the determination state.

[0062] Specifically, the preset height range refers to a pre-defined height interval in the vertical direction (Y-axis) used to determine whether the kicking foot has been effectively retracted from its highest point after the kick. This range is usually set near the height of the starting point of the movement. For example, it can be a range with a tolerance threshold floating up or down based on the height of the starting point of the movement, or it can be set as an absolute height range close to the ground.

[0063] This step is for the final confirmation of the completion of the "falling back" action. The system continuously monitors the height of the moving foot during the falling process. When the foot falls within a preset height range, a complete "lift-sweep-fall" action cycle is considered complete. This ensures the integrity of the action: only after the "retracting" action is completed is the entire intentional action considered finished, further filtering out unnatural actions such as "kicking without retracting" or rapid withdrawal. It also marks the end of the temporal recognition: entering the judgment state means that the temporal state transition based on real-time frame data has been completed, and the system can perform final spatial verification based on the key points recorded throughout the process. This step, together with the previous steps, constitutes a complete and closed real-time action recognition logic chain from start to finish, ensuring that only after the entire prescribed action sequence is completed will the final judgment be entered, thus greatly improving reliability at the temporal level.

[0064] In one optional implementation of this embodiment, Figure 3 This is a schematic diagram of a kicking motion state machine model according to an embodiment of this application. Figure 3As shown, the kicking motion state machine model includes the following states in sequence: initial state A, ready state B, raised state C, sweeping state D, falling state E, and judgment state F. The kicking motion is identified by sequentially transitioning through each state from A to F. If the transition between any two states (B, C, D, E, F) fails to meet the judgment condition, the current state is reset to the initial state A. This achieves phased, interpretable, and temporally sequential tracking and verification of the kicking motion. The design decomposes continuous foot movements into discrete state nodes with clear physical meanings, and each state transition must satisfy preset spatiotemporal conditions. If a transition between any states fails (e.g., the condition is not met), the model immediately resets to the initial state A, enabling real-time interruption and filtering of incomplete, non-standard, or interfering actions. This mechanism not only significantly improves the accuracy and reliability of motion recognition but also fundamentally reduces the possibility of false triggers through the complete flow of the state sequence and abnormal reset logic, thus achieving a balance between high recognition rate and high robustness in complex real-world scenarios.

[0065] In one possible implementation, Figure 4 This is a schematic diagram of a kicking motion trajectory according to an embodiment of this application. Figure 4 As shown, if the currently identified active foot is the right foot, then the left foot is the currently identified standing foot. The kicking motion is identified based on key points recorded during the right foot's kicking process. In this diagram, four points are recorded: A (starting point), B (highest point of elevation), C (farthest sweep), and D. By sequentially identifying and transitioning states based on the key trajectory points ABCD during the kicking process, as indicated by the arrows, the kicking motion of the active foot is identified. Similarly, if the currently identified active foot is the left foot and the right foot is the currently identified standing foot, the kicking motion of the left foot is identified using the same processing method through a state machine. This achieves accurate and stable identification of kicking motions. This method has good versatility and can be applied to both left and right foot motion recognition, providing reliable technical support for applications such as real-time interaction and motion analysis.

[0066] Step S204: When the state of the kicking action state machine transitions to the determination state, a set of spatial key points recorded during the state transition is obtained. The set of spatial key points includes at least the highest point of the raised foot in the lifting state and the farthest point of the sweeping foot in the sweeping state.

[0067] Please see details Figure 1 Step S104 of the illustrated embodiment will not be described again here.

[0068] Step S205: If the spatial geometric relationship between a set of spatial key points satisfies the preset constraint conditions, then it is determined that a valid kicking action has occurred, and the tailgate of the vehicle is opened.

[0069] Specifically, in step S205 above, if the spatial geometric relationship between a set of spatial key points satisfies the preset constraint conditions, then determining that a valid kicking action has occurred includes the following steps: Step S2051: Calculate the vertical displacement of the highest point of lifting relative to the starting point of the action.

[0070] Specifically, based on the three-dimensional spatial coordinates of the starting point of the movement (point A) and the highest point of lifting (point B), the coordinate difference between the two in the vertical direction (usually the Y-axis) is calculated to obtain the vertical displacement ΔY: ΔY = - ,in, To raise the highest point, This represents the height value of the starting point of the action.

[0071] This step quantifies the longitudinal movement amplitude of the foot during the lift-up phase. By precisely quantifying the height of the foot lift-up, an objective and measurable basis is provided for judging whether the action has a sufficient "lift-up" intention, effectively filtering out invalid movements such as low-amplitude leg swaying or ground sliding, and improving the anti-interference ability of the recognition.

[0072] Step S2052: Calculate the horizontal displacement of the furthest point of the sweep relative to the starting point of the action.

[0073] Specifically, based on the three-dimensional spatial coordinates of the starting point of the action (point A) and the farthest point of the sweep (point C), the vector distance between the two in the horizontal direction (Z-axis) is calculated to obtain the horizontal displacement vector ΔZ: ΔZ = - ,in, The z-value of the coordinate point C, which is the furthest point in the sweep. Let z be the z-value of the coordinate point A, the starting point of the action.

[0074] By quantifying the amplitude and direction of foot movement in the horizontal direction, the trajectory characteristics of the "sweeping" motion were accurately captured. Combined with directional analysis, non-target movements such as forward kicks and lateral movements can be distinguished, ensuring that only effective sweeping motions pointing towards the rear of the vehicle are considered.

[0075] Step S2053: If the vertical displacement and horizontal displacement meet the preset constraints, it is determined that a valid kicking action has occurred. The preset constraints include that the vertical displacement is greater than zero and the direction of the horizontal displacement is from the rear of the vehicle to the front of the vehicle.

[0076] Specifically, the calculation results of steps S2051 and S2052 are logically ANDed with preset thresholds and direction conditions. That is, ΔY is required to be greater than Hmin, where Hmin is a threshold greater than zero, and the component of the horizontal displacement ΔZ in the direction from the rear of the vehicle to the front of the vehicle is greater than Lmin, where Lmin is a threshold greater than zero and is the dominant direction.

[0077] By integrating two key spatial geometric features—"sufficient upward lifting" and "clear horizontal sweep towards the rear of the vehicle"—a rigorous dual-criteria system was constructed. This joint judgment mechanism greatly improves the accuracy and reliability of kicking motion recognition, minimizing the probability of false triggering (such as pedestrians passing by) and missed triggering, thereby ensuring that only genuine intentional actions that conform to the preset spatial trajectory features can successfully trigger the tailgate opening.

[0078] In some optional implementations, step S2033 above includes: Step a1: Obtain the reference space coordinates of the starting point of the action.

[0079] Step a2: Calculate the real-time vertical displacement of the moving foot relative to the starting point of the movement based on the spatial coordinates and the reference spatial coordinates.

[0080] Step a3: If the real-time vertical displacement is greater than the first threshold at the first moment, and the vertical movement speed of the moving foot in the motion feature data is less than the first speed threshold, then the displacement of the spatial coordinates in the vertical direction satisfies the first condition, and the three-dimensional coordinates of the spatial coordinates at the first moment are recorded as the highest point of lifting.

[0081] Specifically, the reference spatial coordinates refer to the three-dimensional spatial coordinates of the starting point of the movement. Real-time vertical displacement refers to the height difference in the vertical direction between the current spatial coordinates of the moving foot and the starting point of the movement at any given moment during the lifting process. The first moment refers to a specific point in time where the system simultaneously meets two conditions: the real-time vertical displacement is greater than a first threshold, and the foot's vertical movement speed is less than a first speed threshold. This moment represents the instant the system determines that "the foot has lifted to near its highest point and is approaching stillness." The first threshold is a preset minimum vertical displacement value used to determine whether the foot has performed a sufficiently large lifting motion. The first speed threshold is a preset speed value close to 0 used to determine whether the foot's vertical movement has nearly stopped, i.e., whether it has reached the vicinity of the vertex (highest point) of the parabolic trajectory.

[0082] This implementation provides a precise and robust method for detecting the "highest point of lift." It employs a dual-judgment system combining "height criteria" and "velocity criteria." The vertical displacement of the foot relative to the starting point is continuously monitored; a valid lift is only considered to have occurred when the displacement exceeds a minimum threshold (a first threshold), eliminating minor interference. In addition to meeting the height requirement, the vertical velocity of the foot is simultaneously detected. When the velocity drops below a very low value (a first velocity threshold), it indicates that the foot has reached its peak and is about to fall. The system determines this as the "highest point of lift" at the first moment when both conditions are simultaneously met (the first moment) and records the three-dimensional coordinates of the foot at that moment as point B. This design simulates the intuitive human judgment of the "highest point": it must not only be higher than the starting point but also have a momentary "pause" there.

[0083] This implementation method, by combining vertical displacement thresholds and vertical velocity thresholds, accurately determines and records the highest point of a kicking motion. Real-time vertical displacement is calculated with the starting point of the motion as a reference, ensuring the relativity and stability of the measurement. The highest point is confirmed only when the displacement reaches a certain height and the foot's vertical velocity slows down, which aligns with the physical characteristic that the instantaneous velocity is minimum at the highest point of the kicking motion. This method effectively avoids false detections caused by rapid shaking or unintentional leg swings, enhancing the anti-interference capability and accuracy of kicking state recognition, thus providing reliable key point data for subsequent motion determination.

[0084] In some optional implementations, step S2034 above includes: Step b1: Obtain the reference space coordinates of the starting point of the action.

[0085] Step b2: Calculate the real-time horizontal displacement of the moving foot relative to the starting point of the movement based on the spatial coordinates and the reference spatial coordinates.

[0086] Step b3: If the real-time horizontal displacement is greater than the second threshold at the second moment, and the horizontal movement speed of the moving foot in the motion feature data is less than the second speed threshold, then the displacement of the spatial coordinates in the horizontal direction satisfies the second condition, and the three-dimensional coordinates of the spatial coordinates at the second moment are recorded as the farthest point of the sweep.

[0087] Specifically, real-time horizontal displacement refers to the real-time distance change of the moving foot relative to the starting point of the movement in the horizontal plane during the sweeping motion, used to quantify the amplitude of the foot's backward or lateral movement. The second moment refers to the moment when both conditions are met simultaneously: "real-time horizontal displacement is greater than the second threshold" and "horizontal movement speed is less than the second speed threshold." This moment usually corresponds to the critical point where the foot reaches its farthest position in the horizontal direction and is about to retract. The second threshold is a preset minimum effective sweeping distance used to exclude small-amplitude invalid movements; the second speed threshold is a lower upper limit of the foot's horizontal movement speed, used to determine whether the foot is approaching stillness in the horizontal direction (i.e., reaching its farthest point).

[0088] Similar to steps a1 to a4 above, this implementation accurately determines and records the furthest point of the sweep by jointly judging whether the horizontal displacement is large enough and whether the horizontal movement speed has significantly decreased. This design ensures that the action has a sufficient horizontal movement amplitude and utilizes kinematic laws (the speed is minimum at the furthest point), thereby accurately capturing the turning point of the sweeping action. This enhances the stability and anti-interference ability of sweeping state recognition, effectively avoids false recordings caused by continuous sliding or small swings, and provides reliable key point input for subsequent action validity determination based on spatial geometric relationships.

[0089] In some alternative implementations, the method for opening the vehicle further includes: When the kicking motion state machine model is in any of the states of lifting, sweeping, and falling, if any of the following situations are detected based on the current frame of the image sequence, the kicking motion state machine model will be reset to the initial state: The spatial coordinates of the moving foot were detected to be outside the preset effective detection area. The duration of the current state exceeds the maximum time threshold preset for the current state; Displacement of the standing foot exceeding the stability threshold was detected.

[0090] Specifically, the effective detection area refers to a pre-defined three-dimensional space behind the vehicle, usually located near the rear bumper. Only leg movements within this area will be included in the kicking action recognition process, aiming to focus recognition on the reasonable tailgate operation space. The maximum time threshold refers to the longest allowable duration set for each state such as lifting, sweeping, and lowering. This threshold is set based on ergonomics and the speed of normal kicking actions to prevent the system from "getting stuck" in a certain state due to abnormal conditions. The stability threshold refers to the upper limit of displacement for determining whether the standing foot has moved unexpectedly. When the displacement of the standing foot exceeds this threshold, the body posture is considered unstable and does not meet the prerequisite for kicking while standing on one leg.

[0091] In the core state transition process of motion recognition, a continuous monitoring mechanism based on three dimensions—spatial coordinates, temporal logic, and body stability—is introduced simultaneously. Once any of the three abnormal situations is detected—the moving foot leaving the effective area, any state processing timeout, or significant movement of the standing foot—the state machine is immediately forcibly reset to the initial state, interrupting the current recognition process.

[0092] By adding these three layers of real-time protection rules, the robustness, security, and anti-interference capabilities of the entire recognition system are greatly enhanced. It effectively filters out numerous non-intentional scenarios, such as pedestrians passing behind vehicles, users briefly lingering within the recognition area, or leaving before completing a kicking motion, preventing the system from making misjudgments or remaining in an uncertain intermediate state for extended periods. This ensures that the system completes recognition only under strict spatiotemporal and posture constraints, significantly improving user experience and system reliability.

[0093] In some alternative implementations, after determining that a valid kicking action has occurred in step S205 above, the method further includes: Reset the current state of the kicking action state machine model to the initial state.

[0094] Specifically, in step S205, after the system determines that a valid kicking action has occurred and issues a control command to open the tailgate, it immediately forces the internal state variables of the kicking action state machine model to the initial state, thereby clearing all intermediate process data and restoring the state machine to a standby state where it can start recognizing the next action again.

[0095] First, it ensures that each kicking action recognition process is independent and complete, avoiding interference from residual states of previous actions and preventing continuous false triggers or state logic confusion. Second, through active reset, the system can immediately respond to the next possible legitimate kicking action, improving the continuity of interaction and user experience. Finally, this design enhances the system's robustness, enabling it to quickly return to a definite and stable initial point after processing a clear result, ensuring the continuous and reliable operation of kicking action recognition.

[0096] Figure 5 This is a schematic diagram illustrating a kicking motion state machine model for recognizing kicking motions according to an embodiment of this application. Figure 5As shown, the kicking motion state machine model starts from the initial state (A), detects the user entering the camera's field of view, and sequentially passes through the preparation state (B) where the user begins to kick and records the starting point of the action, the lifting state (C) where the foot is raised to its highest point, the sweeping state (D) where the foot is swept to its furthest point, and the falling state (E) where the kicking action is retracted, finally entering the judgment state (F) to complete a complete motion trajectory acquisition. Only when the moving foot is within the preset effective detection area, the state transitions are triggered continuously in this order, and each state satisfies its corresponding spatial, temporal, and stability conditions, will the state machine reach the judgment state. Then, the validity of the recorded key points is analyzed, thereby achieving reliable and sequential recognition of the kicking action.

[0097] Figure 6 This is a schematic diagram of the state transition of a kicking action state machine model according to an embodiment of this application. Figure 6 As shown, the kicking motion state machine model clearly defines the transition logic and conditions between each state. The recognition process begins in the initial state. When the relative stillness time of both feet exceeds the threshold T0, the system enters the preparation state. In the preparation state, the system records the starting point P(A) of the kicking foot's motion and continuously monitors it. If a valid lifting motion is detected and the displacement of the standing foot does not exceed the threshold T5, the system enters the lifting state and records the highest lifting point P(B).

[0098] Subsequently, when the moving foot moves laterally beyond the threshold T2, the state transitions to a sweeping state. In this state, the system continuously detects and eventually locks onto the furthest sweeping point P(C). When the moving foot sweeps in the opposite direction beyond the threshold T3, i.e., when the sweeping action reaches its furthest point and begins to retract, the state enters a fallback state. In the fallback state, the system records and updates the furthest sweeping point P(C) of the moving foot. Finally, the state enters a judgment state, where the recorded starting point P(A), the highest point of lifting P(B), and the furthest sweeping point P(C) are checked for spatial geometric relationships. If all three meet preset conditions, the "kicking" action is judged as successful. Otherwise, the "kicking" action recognition is judged as failed.

[0099] It is worth noting that in any of the states of preparation, lifting, sweeping, and falling, as long as the movement of the standing foot exceeds the stability threshold T5, or the duration of the movement in the current state exceeds the threshold T6, the state machine will immediately reset to the initial state, thereby effectively eliminating interference and ensuring the accuracy and robustness of action recognition.

[0100] This embodiment also provides a tailgate opening device for a vehicle, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0101] This embodiment provides a tailgate opening device for a vehicle, such as... Figure 7 As shown, it includes: The acquisition module 701 is used to acquire image sequences of the user's feet through an onboard vision sensor. The determination module 702 is used to determine the spatial coordinates and motion feature data of the user's moving foot and standing foot at the current moment based on the current frame of the image sequence; The judgment module 703 is used to determine whether to trigger the transition from the current state to the next state based on the spatial coordinates and motion feature data through the kicking motion state machine model. The kicking motion state machine model predefines multiple node states, which include at least the initial state, the ready state, the raised state, the sweeping state, the falling state, and the judgment state. The acquisition module 704 is used to acquire a set of spatial key points recorded during the state transition when the state machine of the kicking action transitions to the judgment state. The set of spatial key points includes at least the starting point of the movement of the foot in the ready state, the highest point of the foot in the lifting state, and the farthest point of the sweeping state. The tailgate opening module 705 is used to determine that a valid kicking action has occurred and to control the opening of the vehicle's tailgate if the spatial geometric relationship between a set of spatial key points meets the preset constraints.

[0102] In some optional implementations, the determination module 703 includes: The first transfer unit is used to transfer to the preparation state in the initial state when the position of the moving foot is detected to be within a preset effective area and remains stationary. The second transfer unit is used to, in the ready state, when a change in the position of the moving foot is detected, record the current position as the starting point of the action and transfer to the raised state; The third transfer unit is used to record or update the highest point of lifting when the vertical displacement of the moving foot in the lifting state meets the first condition, and then transfer to the sweeping state. The fourth transfer unit is used to record or update the furthest point of the horizontal sweep when the horizontal displacement of the moving foot meets the second condition in the sweep state, and then transfer to the fall state. The fifth transfer unit is used to transfer to the judgment state when the moving foot is detected to fall within a preset height range during the fall state.

[0103] In some alternative implementations, the third transfer unit includes: The position acquisition subunit is used to obtain the reference space coordinates of the action start point; The displacement calculation subunit is used to calculate the real-time vertical displacement of the moving foot relative to the starting point of the action based on the spatial coordinates and the reference spatial coordinates. The lift determination subunit is used to determine that the displacement of the spatial coordinates in the vertical direction satisfies the first condition if the real-time vertical displacement is greater than the first threshold at the first moment and the vertical movement speed of the moving foot in the motion feature data is less than the first speed threshold, and to record the three-dimensional coordinates of the spatial coordinates at the first moment as the highest lift point.

[0104] In some alternative implementations, the fourth transfer unit includes: The position acquisition subunit is used to obtain the reference space coordinates of the action start point; The displacement calculation subunit is used to calculate the real-time horizontal displacement of the moving foot relative to the starting point of the action based on the spatial coordinates and the reference spatial coordinates. The sweep determination subunit is used to determine that the displacement of the spatial coordinates in the horizontal direction satisfies the second condition if the real-time horizontal displacement is greater than the second threshold at the second time moment and the moving foot in the motion feature data moves at a speed less than the second speed threshold in the horizontal direction. The three-dimensional coordinates of the spatial coordinates at the second time moment are recorded as the farthest point of the sweep.

[0105] In some alternative implementations, the tailgate opening module 705 includes: The vertical displacement calculation unit is used to calculate the vertical displacement of the highest point of lifting relative to the starting point of the action; The horizontal displacement calculation unit is used to calculate the horizontal displacement of the furthest point of the sweep relative to the starting point of the action. The effective kicking action determination unit is used to determine that an effective kicking action has occurred if the vertical displacement and horizontal displacement meet preset constraints. The preset constraints include that the vertical displacement is greater than zero and the direction of the horizontal displacement is towards the rear of the vehicle.

[0106] In some alternative embodiments, the apparatus further includes: The reset module is used to reset the kicking motion state machine model to its initial state if any of the following conditions are detected in the current frame of the image sequence when the kicking motion state machine model is in any of the three states: raised, swept, or fallen. The spatial coordinates of the moving foot were detected to be outside the preset effective detection area. The duration of the current state exceeds the maximum time threshold preset for the current state; Displacement of the standing foot exceeding the stability threshold was detected.

[0107] In some alternative implementations, the tailgate opening module 705 includes: The state reset unit is used to reset the current state of the kicking action state machine model to the initial state.

[0108] The tailgate opening device for a vehicle provided in this application can execute the tailgate opening method for a vehicle provided in any embodiment of this application, and has the corresponding functional modules and beneficial effects for executing the method. Further functional descriptions of the various modules and units described above are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0109] Figure 8 This is a schematic diagram illustrating the configuration of a vehicle tailgate opening system according to an embodiment of this application. Figure 8 As shown, the system includes an in-vehicle image acquisition device, an in-vehicle data bus, an image processing and decision-making unit, a body control unit, and an execution structure. Specifically, the execution structure is a tailgate actuator. The in-vehicle data bus connects the in-vehicle image acquisition device and the image processing and decision-making unit. The in-vehicle data bus also connects the image processing and decision-making unit, the body control unit, and the tailgate actuator. The in-vehicle image acquisition device is used to acquire image sequences of the user's feet. The body control unit controls the tailgate actuator to open the tailgate by triggering the action actuator, i.e., the user kicks at the rear of the vehicle. The image processing and decision-making unit is configured to implement the tailgate opening method of the vehicle shown in the above embodiment, recognize the kicking action, and if the recognition is successful, control the tailgate actuator to open the vehicle's tailgate.

[0110] Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0111] The following is a detailed reference. Figure 9 This diagram illustrates a suitable structural schematic for implementing the electronic device described in the embodiments of this application. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 901, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 902 or a program loaded from memory 909 into random access memory (RAM) 903. The RAM 903 also stores various programs and data required for the operation of the electronic device. The processor 901, ROM 902, and RAM 903 are interconnected via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0112] Typically, the following devices can be connected to I / O interface 905: input devices 906 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 907 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 909 including, for example, magnetic tapes, hard disks, etc.; and communication devices 909. Communication device 909 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 9 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0113] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 909, or installed from a memory 909, or installed from a ROM 902. When the computer program is executed by the processor 901, it performs the functions defined in the vehicle tailgate opening method of embodiments of this application.

[0114] Figure 9 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0115] This application also provides a computer-readable storage medium. The methods described in this application can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the vehicle tailgate opening method shown in the above embodiments is implemented.

[0116] A portion of this application can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to this application through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0117] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for opening the tailgate of a vehicle, characterized in that, The method includes: The system uses an in-vehicle vision sensor to capture image sequences of the user's feet. Based on the current frame of the image sequence, determine the spatial coordinates and motion feature data of the user's moving foot and standing foot at the current moment; Based on the spatial coordinates and the motion feature data, a kicking motion state machine model is used to determine whether a transition from the current state to the next state is triggered. The kicking motion state machine model predefines multiple node states, which include at least the initial state, the ready state, the raised state, the sweeping state, the falling state, and the judgment state. When the state of the kicking action state machine transitions to the determination state, a set of spatial key points recorded during the state transition is obtained. The set of spatial key points includes at least the starting point of the movement of the foot in the preparation state, the highest point of the foot in the lifting state, and the farthest point of the foot in the sweeping state. If the spatial geometric relationship between the set of spatial key points satisfies the preset constraints, then a valid kicking action is determined to have occurred, and the tailgate of the vehicle is opened.

2. The method according to claim 1, characterized in that, The step of determining whether to trigger a transition from the current state to the next state based on the spatial coordinates and the motion feature data using a kicking motion state machine model includes: In the initial state, when the position of the moving foot is detected to be within a preset effective area and remains stationary, the process transitions to the preparation state. In the preparation state, when a change in the position of the moving foot is detected, the current position is recorded as the starting point of the action, and the process transitions to the raised state. In the raised state, when the vertical displacement of the moving foot meets the first condition, the highest point of the raised position is recorded or updated, and the process transitions to the sweeping state. In the sweeping state, when the horizontal displacement of the moving foot meets the second condition, the furthest point of the sweep is recorded or updated, and the process transitions to the falling state. In the falling state, when the moving foot is detected to have fallen to a preset height range, the process transitions to the determination state.

3. The method according to claim 2, characterized in that, When the vertical displacement of the moving foot is detected to meet the first condition, the highest point of lifting is recorded or updated, including: Obtain the reference space coordinates of the starting point of the action; Calculate the real-time vertical displacement of the moving foot relative to the starting point of the action based on the spatial coordinates and the reference spatial coordinates; If the real-time vertical displacement is greater than the first threshold at the first moment, and the vertical movement speed of the moving foot in the motion feature data is less than the first speed threshold, it is determined that the vertical displacement of the spatial coordinates satisfies the first condition, and the three-dimensional coordinates of the spatial coordinates at the first moment are recorded as the highest point of lifting.

4. The method according to claim 2, characterized in that, When the displacement of the spatial coordinates in the horizontal direction is detected to meet the second condition, the process of recording or updating the furthest point of the sweep includes: Obtain the reference space coordinates of the starting point of the action; Calculate the real-time horizontal displacement of the moving foot relative to the starting point of the action based on the spatial coordinates and the reference spatial coordinates; If the real-time horizontal displacement is greater than the second threshold at the second moment, and the horizontal movement speed of the moving foot in the motion feature data is less than the second speed threshold, it is determined that the displacement of the spatial coordinate in the horizontal direction satisfies the second condition, and the three-dimensional coordinate of the spatial coordinate at the second moment is recorded as the farthest point of the sweep.

5. The method according to claim 1, characterized in that, If the spatial geometric relationship between the set of spatial key points satisfies the preset constraint conditions, then a valid kicking action is determined to have occurred, including: Calculate the vertical displacement of the highest point of lifting relative to the starting point of the action; Calculate the horizontal displacement of the furthest point of the sweep relative to the starting point of the action; If the vertical displacement and the horizontal displacement satisfy the preset constraint conditions, then a valid kicking action is determined to have occurred, wherein the preset constraint conditions include the vertical displacement being greater than zero and the horizontal displacement being directed toward the rear of the vehicle.

6. The method according to claim 1, characterized in that, The method further includes: When the kicking motion state machine model is in any of the raised state, the sweeping state, and the falling state, if any of the following situations are detected based on the current frame of the image sequence, the kicking motion state machine model is reset to the initial state: The spatial coordinates of the moving foot were detected to be outside the preset effective detection area. The duration of the current state exceeds the maximum time threshold preset for the current state; The displacement of the standing foot exceeding the stability threshold was detected.

7. The method according to claim 1, characterized in that, After determining that a valid kick has occurred, the method further includes: The current state of the kicking motion state machine model is reset to the initial state.

8. A tailgate opening device for a vehicle, characterized in that, The device includes: The acquisition module is used to acquire image sequences of the user's feet through an onboard vision sensor; The determination module is used to determine the spatial coordinates and motion feature data of the user's moving foot and standing foot at the current moment based on the current frame of the image sequence; The judgment module is used to determine whether to trigger a transition from the current state to the next state based on the spatial coordinates and the motion feature data through a kicking motion state machine model. The kicking motion state machine model predefines multiple node states, which include at least the initial state, the ready state, the raised state, the sweeping state, the falling state, and the judgment state. The acquisition module is used to acquire a set of spatial key points recorded during the state transition when the state of the kicking action state machine transitions to the determination state. The set of spatial key points includes at least the starting point of the movement of the foot in the preparation state, the highest point of the foot in the lifting state, and the farthest point of the sweeping motion of the foot in the sweeping state. The tailgate opening module is used to determine that a valid kicking action has occurred and to control the opening of the vehicle's tailgate if the spatial geometric relationship between the set of spatial key points meets the preset constraints.

9. A tailgate opening system for a vehicle, characterized in that, The system includes an in-vehicle image acquisition device, an in-vehicle data bus, an image processing and decision-making unit, a body control unit, and a tailgate actuator. The in-vehicle data bus connects the in-vehicle image acquisition device and the image processing and decision-making unit, and also connects the image processing and decision-making unit, the body control unit, and the tailgate actuator. The in-vehicle image acquisition device is used to acquire image sequences of the user's feet. The body control unit controls the tailgate actuator to open the tailgate. The image processing and decision-making unit is configured to perform the tailgate opening method of the vehicle as described in any one of claims 1 to 7.

10. A vehicle, characterized in that, The vehicle includes a controller, which includes a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to perform the tailgate opening method of the vehicle according to any one of claims 1 to 7.

Images