Parking attitude self-adaptive adjustment method and system based on vehicle micro-motion feature recognition

Abstract

The application relates to a parking posture self-adaptive adjustment method and system based on vehicle micro-motion feature recognition, a computer device and a storage medium. The method comprises the following steps: constructing a micro-motion feature vector according to micro-motion features, the micro-motion features comprising a steering wheel angle change rate and a steering delay time; identifying a driver operation mode based on micro-motion feature vector statistical characteristics; simultaneously determining a sampling strategy, an expansion step and a steering rate constraint of a quick expansion random tree star algorithm and PID parameters of a PID control algorithm according to the driver operation mode; running the algorithm to generate a parking trajectory and extract an expected steering angle; predicting the steering angle by using Kalman filtering and calculating a deviation of the steering angle from the expected steering angle; when the deviation exceeds a preset threshold, outputting a corrected steering angle by using a PID algorithm and converting the corrected steering angle into a direction correction prompt; collecting a micro-motion feature vector adjusted by a driver and cyclically executing until the vehicle reaches a standard parking posture. The application improves the accuracy of parking guidance and the utilization rate of parking spaces.

Description

Technical Field

[0001] This application relates to the field of intelligent parking assistance technology, and in particular to a parking posture adaptive adjustment method, system, computer equipment and storage medium based on vehicle micro-motion feature recognition. Background Technology

[0002] With the continuous growth of car ownership, parking space is becoming increasingly limited, making it more difficult for drivers to park in narrow spaces. Smart parking systems, as an important component of urban static traffic management, improve parking lot management efficiency and user experience through functions such as parking guidance, reservation parking, and contactless payment. However, the service scope of existing smart parking systems is mainly focused on parking space resource scheduling and information services, and has not yet extended to real-time assistance during the vehicle parking process.

[0003] In current smart parking scenarios, parking lots can obtain real-time vehicle driving data (such as speed and steering angle) through V2X (Vehicle-to-Everything) or 5G networks, providing the technological foundation for sensing and intervening in the parking process. However, existing smart parking systems do not use driver operation characteristics uploaded by the vehicle (such as steering wheel speed and steering response delay) as dynamic input parameters for real-time generation and correction of parking trajectories. This results in the system being unable to provide differentiated parking guidance services based on the driver's individual operating habits, forcing drivers to rely entirely on their own experience to make directional adjustments during parking. When improper operation by the driver results in repeated adjustments that fail to achieve a standard parking position, the vehicle occupies passageway space for an extended period, causing queuing congestion for subsequent vehicles; or, due to incorrect vehicle posture, it occupies adjacent parking spaces, reducing the actual number of available parking spaces. Especially during peak hours, these issues reduce parking lot turnover efficiency and weaken the actual effectiveness of smart parking systems in resource scheduling. Summary of the Invention

[0004] Therefore, it is necessary to provide a parking posture adaptive adjustment method, system, computer equipment, and storage medium based on vehicle micro-motion feature recognition, which can adaptively generate parking trajectory and direction correction prompts according to the micro-motion features of driver operation, and guide the driver to complete parking in a standard posture.

[0005] Firstly, this application provides a parking posture adaptive adjustment method based on vehicle micro-motion feature recognition. The method includes: S1: Construct a micro-motion feature vector based on the micro-motion characteristics; the micro-motion characteristics include the steering wheel angle change rate and steering delay time; S2: Identify the driver's operating mode based on the statistical characteristics of the micro-motion feature vector at the current moment. Using the driver's operating mode as the same input, simultaneously determine the sampling strategy, expansion step size, and steering rate constraints of the Fast Extended Random Tree Star Algorithm, as well as the PID parameters of the PID control algorithm, so that the parking trajectory generated by the Fast Extended Random Tree Star Algorithm and the corrected steering angle output by the PID control algorithm are both adapted to the driver's operating mode. Run the Fast Extended Random Tree Star Algorithm with the sampling strategy, the expansion step size, and the steering rate constraints to generate a parking trajectory, and extract the expected steering angle for each time step from the parking trajectory. S3: Using Kalman filtering, predict the steering angle for the next N time steps based on the micro-motion feature vector, and calculate the prediction deviation between the predicted steering angle and the desired steering angle; when the absolute value of the prediction deviation exceeds a preset deviation threshold, input the prediction deviation into the PID control algorithm, the PID control algorithm outputs a corrected steering angle with the PID parameters, and converts the corrected steering angle into a direction correction prompt output to the driver; S4: Collect the micro-motion feature vector after the driver's adjustment operation, use it as the new micro-motion feature vector, and return to S2 until the vehicle reaches the standard parking posture.

[0006] In one embodiment, identifying the driver's operating pattern based on the statistical characteristics of the micro-motion feature vector includes: The standard deviation of the steering wheel angle change rate and the average value of the steering delay time are calculated based on the micro-motion feature vector at multiple consecutive time points. The standard deviation of the steering wheel angle change rate is compared with a preset standard deviation threshold, and the average value of the steering delay time is compared with a preset average value threshold. When the standard deviation of the steering wheel angle change rate is less than the standard deviation threshold and the average value of the steering delay time is greater than the average value threshold, the driver operation mode is identified as a conservative mode; otherwise, the driver operation mode is identified as an aggressive mode.

[0007] In one embodiment, determining the sampling strategy, the extended step size, and the steering ratio constraint based on the driver's operating mode includes: When the driver's operation mode is conservative, the sampling strategy is to randomly sample nodes preferentially distributed in the smooth path area, the expansion step size is the first expansion step size, and the steering rate constraint is that the steering wheel angle change rate does not exceed the preset angle change rate threshold. When the driver's operation mode is the aggressive mode, the sampling strategy is that the distribution range of the random sampling nodes is greater than the distribution range in the conservative mode, the expansion step size is a second expansion step size greater than the first expansion step size, and the steering rate constraint is that the steering wheel angle change rate does not exceed the sum of the angle change rate threshold and the mode coefficient.

[0008] In one embodiment, the PID parameters include proportional gain, integral gain, and derivative gain; determining the PID parameters according to the driver's operating mode includes: When the driver's operating mode is the conservative mode, the corresponding proportional gain is the first proportional gain value, the corresponding integral gain is the first integral gain value, and the corresponding derivative gain is the first derivative gain value. When the driver's operating mode is the aggressive mode, the corresponding proportional gain is a second proportional gain value that is greater than the first proportional gain value, the corresponding integral gain is a second integral gain value that is less than the first integral gain value, and the corresponding derivative gain is a second derivative gain value that is greater than the first derivative gain value.

[0009] In one embodiment, the micro-motion feature vector further includes steering wheel angle, vehicle speed, and wheel trajectory curvature radius, wherein the steering wheel angle change rate is calculated based on the collected steering wheel angle, and the wheel trajectory curvature radius is calculated based on the steering wheel angle and the vehicle speed.

[0010] In one embodiment, the radius of curvature of the wheel track is calculated using the following formula: in, For the vehicle speed, This refers to the vehicle's wheelbase. The front wheel steering angle is determined based on the steering wheel angle and steering ratio.

[0011] In one embodiment, the dynamic constraints that the node expansion of the fast expanding random tree star algorithm must satisfy during the generation of the parking trajectory also include steering wheel angle constraints and radius of curvature constraints. The steering wheel angle constraint is that the absolute value of the steering wheel angle does not exceed the maximum steering wheel angle set according to the vehicle type; the radius of curvature constraint is that the radius of curvature of the wheel trajectory is not less than... ,in, For vehicle speed, The coefficient of friction of the road surface. This is the acceleration due to gravity.

[0012] Secondly, this application also provides a parking posture adaptive adjustment system based on vehicle micro-motion feature recognition. The system includes: The micro-motion feature construction module is used to construct a micro-motion feature vector based on micro-motion features; the micro-motion features include the steering wheel angle change rate and steering delay time. The trajectory planning module is used to identify the driver's operating mode based on the statistical characteristics of the micro-motion feature vector at the current moment. Using the driver's operating mode as the same input, it simultaneously determines the sampling strategy, expansion step size, and steering rate constraints of the Fast Extended Random Tree Star Algorithm (FRADA), as well as the PID parameters of the PID control algorithm. This ensures that the parking trajectory generated by the FRADA and the corrected steering angle output by the PID control algorithm are both adapted to the driver's operating mode. The FRADA is then run using the sampling strategy, expansion step size, and steering rate constraints to generate the parking trajectory, and the expected steering angle for each time step is extracted from the parking trajectory. The prediction correction module is used to predict the steering angle for the next N time steps based on the micro-motion feature vector using Kalman filtering, and to calculate the prediction deviation between the predicted steering angle and the desired steering angle; when the absolute value of the prediction deviation exceeds a preset deviation threshold, the prediction deviation is input into the PID control algorithm, the PID control algorithm outputs the corrected steering angle with the PID parameters, and converts the corrected steering angle into a direction correction prompt and outputs it to the driver; The interactive feedback module is used to collect the micro-motion feature vector after the driver's adjustment operation, as the new micro-motion feature vector, and trigger the trajectory planning module and the prediction correction module to be executed repeatedly until the vehicle reaches the standard parking posture.

[0013] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the above-described adaptive parking posture adjustment method based on vehicle micro-motion feature recognition.

[0014] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the above-described parking posture adaptive adjustment method based on vehicle micro-motion feature recognition.

[0015] The aforementioned parking posture adaptive adjustment method, system, computer equipment, and storage medium based on vehicle micro-motion feature recognition achieve quantitative characterization of driver operation characteristics by constructing a micro-motion feature vector using the steering wheel angle change rate and steering delay time as micro-motion features. Based on the statistical characteristics of the micro-motion feature vector, the driver's operation mode is identified. This operation mode is then used as the same input to simultaneously determine the sampling strategy, expansion step size, and steering rate constraints of the fast extended stochastic tree star algorithm, as well as the PID parameters of the PID control algorithm. This ensures that the parking trajectory generated by the fast extended stochastic tree star algorithm and the corrected steering angle output by the PID control algorithm are both adapted to the driver's operation mode. This system achieves adaptive matching of parking trajectory planning and direction correction control to driver operation characteristics. It utilizes Kalman filtering to predict the future steering angle based on micro-motion feature vectors and calculates the deviation from the desired steering angle. The PID control algorithm is only triggered to output a corrected steering angle when the deviation exceeds a preset threshold, avoiding frequent corrections that could interfere with driver operation. The corrected steering angle is then converted into a direction correction prompt and output to the driver. The system also collects the micro-motion feature vector after the driver's adjustment and uses it as a new input to repeat the above process, forming a closed-loop human-machine collaborative control system. This allows for continuous tracking of the driver's real-time operation characteristics in each iteration and dynamic adjustment of subsequent trajectory planning and correction strategies. Attached Figure Description

[0016] Figure 1 This is an application environment diagram of a parking posture adaptive adjustment method based on vehicle micro-motion feature recognition in one embodiment; Figure 2 This is a flowchart illustrating a parking posture adaptive adjustment method based on vehicle micro-motion feature recognition in one embodiment. Figure 3 This is a flowchart illustrating the process of identifying driver operation modes in one embodiment; Figure 4 This is a block diagram of a parking posture adaptive adjustment system based on vehicle micro-motion feature recognition in one embodiment. Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. First, to facilitate understanding of the technical solutions provided by the embodiments of this application, the background technology involved in the embodiments of this application will be described below.

[0018] With the continuous growth of car ownership, urban parking space is becoming increasingly limited. On the one hand, the land cost for new parking lots in urban core areas is constantly rising, while the growth rate of parking space supply is far lower than the growth rate of car ownership, resulting in a continuous compression of the area per unit parking space and the frequent breaching of the minimum size of standard parking spaces. On the other hand, the widespread adoption of special parking scenarios such as mechanical multi-level parking garages and underground narrow-lane parking lots places higher demands on the precision of drivers' parking operations. Against this backdrop, the difficulty for drivers to complete parking operations in narrow spaces is constantly increasing, and the problems of traffic congestion and waste of parking resources caused by parking difficulties are becoming increasingly prominent.

[0019] As a crucial component of urban static traffic management, smart parking systems have experienced rapid development in recent years. Existing smart parking systems typically integrate functions such as parking guidance, parking reservation, and contactless payment. Specifically, the parking guidance function uses geomagnetic sensors, parking space cameras, or ultrasonic detectors to monitor the occupancy status of each parking space in real time and pushes information about available spaces to the guidance screen at the parking lot entrance or the user's mobile application, guiding drivers to quickly find available parking spaces. The parking reservation function allows drivers to pre-lock parking spaces via a mobile application before arriving at the parking lot, reducing the time spent searching for a space upon arrival. The contactless payment function uses license plate recognition or electronic tag recognition to automatically deduct parking fees when the vehicle leaves the parking lot, eliminating the need for the driver to stop and pay. These functions improve parking lot management efficiency and user experience to varying degrees.

[0020] However, the service scope of existing smart parking systems is mainly concentrated on parking space resource scheduling and information services, and has not yet extended to real-time assistance in the parking process. Specifically, after the driver arrives at the designated parking space through the parking space guidance function, the existing smart parking system cannot provide any substantial parking operation assistance during the crucial process from the moment the vehicle enters the vicinity of the parking space to when the vehicle is fully parked. During the parking process, the driver still needs to rely entirely on their own experience to complete steering wheel operation, speed control, and vehicle posture adjustment. The system neither senses the driver's operating characteristics nor intervenes in guiding the parking trajectory.

[0021] In current smart parking scenarios, with the gradual maturation of V2X and 5G communication technologies, parking lots now possess the technological foundation for real-time data interaction with vehicles. Specifically, vehicles can collect their own real-time driving data, including speed, steering wheel angle, and gear information, through onboard diagnostic interfaces or vehicle infotainment systems. This data is then uploaded in real-time to edge computing devices or cloud servers at the parking lot via V2X or 5G communication modules. Based on this data, the parking lot can perceive and track the vehicle's driving status and parking progress, thus enabling the technical possibility of intervening in the parking process and providing real-time assistance.

[0022] However, while existing smart parking systems can acquire basic data such as vehicle speed and steering wheel angle, they do not deeply mine and utilize this data. Specifically, a driver's steering wheel operation during parking contains rich personalized characteristics, with significant differences among drivers in steering wheel speed and steering response delay. For example, experienced drivers typically have a faster steering rate and shorter steering delay, resulting in crisp and decisive operation; while inexperienced drivers or those lacking confidence in the current parking space size usually have a slower steering rate and longer steering delay, exhibiting hesitant operation. These differences directly determine the parking trajectory and the vehicle's final parking position. However, existing smart parking systems do not incorporate these driver operation characteristics as dynamic input parameters into the real-time generation and correction of the parking trajectory, preventing the system from providing differentiated parking guidance services based on the driver's individual operating habits.

[0023] Against this technological backdrop, when a driver's improper operation causes the vehicle to repeatedly adjust but still fail to park in a standard position, a series of chain reactions occur. First, the vehicle repeatedly moves forward and backward within the parking lane, occupying lane space for an extended period, preventing waiting vehicles from passing. During peak hours, this can cause severe congestion, disrupting the overall traffic flow in the parking lot. Second, when the vehicle finally parks at an angle, part of the vehicle may extend beyond the parking space boundary, encroaching on adjacent spaces and reducing the number of usable spaces. Given the already limited parking resources, this hidden waste of space due to improper parking further exacerbates the parking problem. These issues reduce the parking lot's turnover rate and weaken the effectiveness of smart parking systems in resource allocation and information services.

[0024] Therefore, in order to identify the driver's personalized operating characteristics in real time and dynamically adjust the parking assistance strategy accordingly, this application provides a parking posture adaptive adjustment method, system, computer device and storage medium based on vehicle micro-motion feature recognition.

[0025] The parking posture adaptive adjustment method based on vehicle micro-motion feature recognition provided in this application can be applied to, for example... Figure 1In the application environment shown, vehicle terminal 102 communicates with server 104 via a network. Vehicle terminal 102 is mounted on the target vehicle and is used to collect real-time driving data (including steering wheel angle, vehicle speed, etc.) and upload the data to server 104. Server 104 uses the real-time driving data uploaded by vehicle terminal 102 to execute the parking posture adaptive adjustment method provided in this embodiment of the application, generating a direction correction prompt. Parking terminal 106 is located in the parking lot and is used to receive the direction correction prompt returned by server 104 and present the direction correction prompt to the driver visually or verbally. Data storage system can store the data that server 104 needs to process. The data storage system can be integrated on server 104 or placed on the cloud or other network servers. Vehicle terminal 102 can be, but is not limited to, an in-vehicle terminal, in-vehicle controller, in-vehicle infotainment system, a driving recorder with communication function, a smart rearview mirror, or an in-vehicle diagnostic interface device. Parking terminal 106 can be, but is not limited to, a display screen near the parking space, an information display screen next to the parking lot passage, a voice broadcast device, or an interactive terminal integrated into the parking management system. Server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers, or it can be deployed on edge computing devices at the parking lot.

[0026] Furthermore, in the specific implementation of the technical solution provided in this application, the data collected by the vehicle terminal 102 is only the steering wheel angle and vehicle speed. This data represents the instantaneous operating state parameters of the vehicle during a single parking process and does not involve continuous tracking of the vehicle's trajectory, nor does it involve sensitive personal information such as audio, video, images, or biometric features. The purpose of data collection is limited to generating parking trajectory and direction correction prompts to assist the driver in completing the parking operation. It is directly related to the aforementioned purpose and represents the minimum scope necessary to achieve that purpose. In practical applications, the purpose, method, and scope of data collection are fully explained to the driver before data collection, and the driver's explicit consent is obtained. The collected data is used only for this parking assistance and not for any other purpose. It is also promptly deleted or anonymized after parking is completed.

[0027] Firstly, during parking assistance, existing parking management systems cannot provide differentiated parking guidance services based on drivers' individual operating habits, forcing drivers to rely entirely on their own experience to make directional adjustments during parking. When drivers operate incorrectly, the vehicle needs repeated adjustments to park, occupying aisle space for an extended period and causing congestion for subsequent vehicles; or, due to incorrect vehicle posture, it may occupy adjacent parking spaces, reducing the number of available parking spaces. The root cause of these problems is that existing parking management systems bind parking trajectory generation to environmental perception data, but fail to use the driver's micro-motion characteristics as dynamic input parameters for trajectory generation. This results in the system-generated trajectory being unable to adapt to the individual operating habits of different drivers, and a lack of effective collaboration mechanism between the driver and the parking management system. Therefore, in one embodiment, such as... Figure 2 As shown, a parking posture adaptive adjustment method based on vehicle micro-motion feature recognition is provided, and this method is applied to... Figure 1 Taking server 104 as an example, the following steps are included: S1: Construct a micro-motion feature vector based on the micro-motion characteristics; the micro-motion characteristics include the steering wheel angle change rate and steering delay time; S2: Identify the driver's operating mode based on the statistical characteristics of the micro-motion feature vector at the current moment. Using the driver's operating mode as the same input, determine the sampling strategy, expansion step size and steering rate constraints of the fast extended stochastic tree star algorithm, as well as the PID parameters of the PID control algorithm, so that the parking trajectory generated by the fast extended stochastic tree star algorithm and the corrected steering angle output by the PID control algorithm are both adapted to the driver's operating mode. Run the fast extended stochastic tree star algorithm with the sampling strategy, expansion step size and steering rate constraints to generate the parking trajectory, and extract the expected steering angle at each time step from the parking trajectory. S3: Using Kalman filtering, predict the steering angle for the next N time steps based on the micro-motion feature vector, and calculate the prediction deviation between the predicted steering angle and the desired steering angle; when the absolute value of the prediction deviation exceeds the preset deviation threshold, the prediction deviation is input into the PID control algorithm, the PID control algorithm outputs the corrected steering angle with PID parameters, and the corrected steering angle is converted into a direction correction prompt output to the driver. S4: Collect the micro-motion feature vector after the driver's adjustment operation, use it as the new micro-motion feature vector, and return to S2 until the vehicle reaches the standard parking posture.

[0028] Specifically, when a vehicle is about to enter the target parking space, server 104 first obtains the location information of the target parking space. The location information of the target parking space is determined by parking terminal 106 through any of the existing technologies such as geomagnetic sensor detection, ultrasonic sensor detection, or video image analysis, and then reported to server 104. Subsequently, server 104 begins to execute the micro-motion feature vector construction step (i.e., S1) of this method. It should be noted that the specific method for determining the target parking space belongs to the prior art, and this application embodiment does not limit it.

[0029] In S1, server 104 receives sensor data uploaded by vehicle terminal 102. This data includes the steering wheel angle collected by the steering wheel angle sensor and the vehicle speed collected by the vehicle speed sensor. The steering wheel angle change rate refers to the amount of change in steering wheel angle per unit time, reflecting the driver's steering wheel rotation speed and operating style, measured in degrees per second (° / s). The steering delay time refers to the time from when the driver begins to turn the steering wheel to when the vehicle steering system responds, reflecting the delay characteristics between driver operation and vehicle response, measured in seconds (s). In practical applications, server 104 receives sensor data uploaded by vehicle terminal 102 via vehicle-to-everything (V2X) communication or 5G communication. This data includes the steering wheel angle collected by the steering wheel angle sensor and the vehicle speed collected by the vehicle speed sensor. The steering wheel angle change rate can be obtained by differential calculation of the received steering wheel angle data. Specifically, steering wheel angle values ​​are received at consecutive time points, and the difference between the steering wheel angles of two adjacent time points is divided by the time interval to obtain the steering wheel angle change rate within that time interval. Steering delay time can be determined by detecting the time difference between the point when the steering wheel angle begins to change and the point when the vehicle yaw rate begins to change. Server 104 combines the calculated current steering wheel angle change rate and the current steering delay time into a micro-motion feature vector, which is used to characterize the driver's operating characteristics at the current moment.

[0030] In S2, the driver's operating mode is first identified based on the statistical characteristics of the micro-motion feature vector at the current moment. During vehicle operation, server 104 continuously receives the steering wheel angle change rate and steering delay time uploaded by vehicle terminal 102, forming a micro-motion feature vector sequence. Server 104 performs statistical analysis on this sequence, calculating the standard deviation of the steering wheel angle change rate and the average value of the steering delay time. Based on these statistical values, it identifies whether the driver operates in a conservative or aggressive mode. A conservative mode indicates a smooth driving style with small changes in steering speed but a relatively slow response, while an aggressive mode indicates a decisive driving style with large changes in steering speed and a rapid response.

[0031] Then, server 104 uses the identified driver operating mode as the same input to simultaneously perform two parameter determination tasks. First, it determines the sampling strategy, expansion step size, and steering rate constraint for the Fast Expanding Random Tree Star Algorithm (FRAS) based on the driver operating mode. The FRAS is an asymptotically optimal path search algorithm based on random sampling, which searches for feasible paths from the starting point to the ending point by progressively expanding a random tree in the workspace. The sampling strategy determines the spatial distribution of randomly sampled nodes generated during the expansion process; the expansion step size determines the distance between the new node and the nearest node during each expansion, in meters (m); and the steering rate constraint limits the upper limit of the rate of change of the steering wheel angle during trajectory tracking, in degrees per second (° / s). Second, it determines the proportional gain, integral gain, and derivative gain of the PID control algorithm based on the driver operating mode. These three gain parameters jointly determine the response characteristics of the PID control algorithm to input deviations. By simultaneously using the driver operating mode to determine the three parameters of both the FRAS and the PID control algorithm, trajectory generation and correction control are adapted to the same driver operating mode.

[0032] After determining the two sets of parameters mentioned above, server 104 runs the Fast Extended Random TreeStar algorithm with the determined sampling strategy, expansion step size, and steering ratio constraints to generate a parking trajectory from the vehicle's current position to the target parking position. This parking trajectory is a continuous path consisting of a series of waypoints, each waypoint containing the vehicle's position coordinates and orientation angle at that location. Server 104 extracts the expected steering angle for each time step from the generated parking trajectory. The expected steering angle refers to the steering wheel angle value required for the vehicle to remain on the trajectory at each waypoint, measured in degrees (°). Specifically, based on the curvature of the waypoints and the vehicle's steering ratio, the path curvature is converted into the corresponding steering wheel angle to obtain the expected steering angle for each time step.

[0033] In S3, Kalman filtering is an optimal estimation algorithm based on a state-space model, which predicts and updates the vehicle's driving state recursively. In this step, server 104 uses the micro-motion feature vector (containing state information such as steering wheel angle and vehicle speed) as input to the Kalman filter, and uses the prediction step of the Kalman filter to predict the steering angle for the next N time steps, obtaining the predicted steering angle, denoted as . subscript This represents the state estimate for the k-th time step based on information from the k-th time step and earlier, expressed in degrees (°). Server 104 compares the predicted steering angle with the expected steering angle for the corresponding time step extracted from S2, denoted as . The subscript k represents the k-th time step, and the unit is degrees (°). Calculate the difference between the two. The prediction deviation is obtained, expressed in degrees (°). The prediction deviation reflects the degree of deviation between the driver's current operating trend and the ideal trajectory.

[0034] After calculating the prediction bias, server 104 determines the absolute value of the prediction bias. Does it exceed the preset deviation threshold? Deviation threshold This is a pre-set angle value, measured in degrees (°), used to determine whether the deviation has reached a level requiring intervention. If the absolute value of the predicted deviation does not exceed the deviation threshold, it indicates that the driver's current operational deviation is within an acceptable range and no intervention is needed; the server 104 continues monitoring without outputting a correction command. Only when the absolute value of the predicted deviation exceeds the preset deviation threshold does the server 104 use the predicted deviation as input to the PID control algorithm. The PID control algorithm performs proportional, integral, and derivative calculations on the predicted deviation using the PID parameters determined in S2, outputting a corrected steering angle. The unit is degrees (°). This mechanism of "intervening only when the deviation exceeds the limit" avoids frequent corrections that would interfere with the driver's normal operation.

[0035] The server 104 will obtain the corrected steering angle The information is converted into a prompt that the driver can understand. The steering angle correction is a specific angular value, indicating the suggested direction and angle of steering wheel rotation for the driver, measured in degrees (°). This angle value is converted into a steering correction prompt, such as a voice prompt saying "Turn the steering wheel X degrees to the left" or "Turn the steering wheel X degrees to the right," or a directional arrow and angle value displayed on a screen. Server 104 sends this steering correction prompt to parking terminal 106. Parking terminal 106 is located within the parking lot and may be, but is not limited to, a display screen near a parking space, an information display screen next to the parking lane, a voice broadcast device, or an interactive terminal integrated into the parking management system. After receiving the steering correction prompt from server 104, parking terminal 106 presents the prompt to the driver visually or audibly, for example, by playing a voice prompt saying "Turn the steering wheel X degrees to the left" or "Turn the steering wheel X degrees to the right" through a loudspeaker at the corresponding parking space, or by displaying a directional arrow and angle value on a screen at the corresponding parking space.

[0036] In S4, server 104 continues to receive steering wheel angle and vehicle speed data after driver adjustments uploaded by vehicle terminal 102. It recalculates the steering wheel angle change rate and steering delay time in the same manner as in S1, and constructs a new micro-motion feature vector as input for the next iteration. Then, it returns to S2, which involves re-identifying the driver's operating mode, re-determining algorithm parameters, regenerating the parking trajectory, re-performing Kalman filter prediction and deviation calculation, and conditionally outputting a new corrected steering angle, which is then converted into a new direction correction prompt. This iterative cycle continues until the vehicle reaches a standard parking posture. A standard parking posture refers to the final state where the vehicle is completely parked in the target parking space, with the vehicle body parallel to the parking space boundary and the vehicle's centerline aligned with the parking space's centerline. The loop terminates when server 104 determines that the vehicle has reached the standard parking posture.

[0037] In an optional implementation of this embodiment, to ensure the real-time performance and feasibility of the above method in practical applications, the rapidly expanding random tree star algorithm is configured with a maximum number of iterations. During algorithm operation, when the maximum number of iterations is reached, the algorithm automatically terminates the path search in the current round and outputs the currently found optimal path. Simultaneously, the algorithm also includes a timeout mechanism; when the execution time of a single path search exceeds a preset time threshold, the algorithm automatically terminates and outputs the currently found optimal path. By setting the maximum number of iterations and the timeout mechanism, excessive computation time in extremely complex scenarios can be avoided, preventing the server 104 from failing to output correction prompts in real time. In practical applications, the maximum number of iterations can be set according to the computing power of the server 104's computing platform. It should be noted that this application embodiment does not limit the specific value of the maximum number of iterations; those skilled in the art can set it reasonably according to the actual application scenario and the capabilities of the computing platform.

[0038] In an optional implementation of this embodiment, the fast-expanding random tree star algorithm employs an incremental computation strategy, reusing the search tree built in the previous round of search in each iteration. Specifically, in the previous iteration, the algorithm has built a random tree from the starting point to the ending point and generated the parking trajectory. When the driver adjusts the operation according to the direction correction prompt, the vehicle's current position and state change, but the difference between the new starting point and the previous starting point is usually small. Therefore, in this iteration, the server 104 does not rebuild the random tree from scratch, but instead, based on retaining the previous search tree, prunes and updates the search tree according to the current vehicle position and continues to expand new nodes. This incremental computation strategy can reduce the computational load of path search in each iteration, further shortening the response time of the server 104. In actual parking scenarios, after adopting the incremental fast-expanding random tree star strategy, the computation time for a single path planning can be controlled within 0.15 seconds.

[0039] In an optional embodiment of this example, the server 104 and the vehicle terminal 102 transmit data via vehicle-to-infrastructure (V2I) communication or 5G communication. The vehicle terminal 102 uploads the collected steering wheel angle and vehicle speed data to the server 104 via V2I or 5G network. The server 104 executes the steps S1 to S4 described above to generate a direction correction prompt and transmits the direction correction prompt back to the parking terminal 106 via V2I or 5G network. Since the parking process is a low-speed driving scenario, the vehicle speed is usually no more than 5 kilometers per hour. The distance the vehicle travels in a single iteration cycle (including uplink data transmission, server 104 calculation, and downlink data transmission) is very short. The transmission delay of V2I or 5G communication (usually 10 milliseconds to 50 milliseconds) is very small compared to the driver's reaction time (usually 0.2 seconds to 1.0 seconds) and will not affect parking safety. It should be noted that the above transmission delay data was obtained based on actual test environments using 5G communication networks (downlink peak rate not less than 100Mbps) and vehicle-to-everything (V2X) communication (communication distance not less than 300 meters), and ranges from 10 milliseconds to 50 milliseconds; the reaction time data was obtained based on actual test results of multiple drivers' operation responses in parking scenarios, and ranges from 0.2 seconds to 1.0 second. Actual values ​​may vary depending on network conditions and individual driver differences, and this application embodiment does not limit these.

[0040] This embodiment constructs a micro-motion feature vector using the steering wheel angle change rate and steering delay time as micro-motion features, thereby achieving a quantitative representation of the driver's operation characteristics. By identifying the driver's operation mode based on the statistical characteristics of the micro-motion feature vector, and using this operation mode as the same input, it simultaneously determines the sampling strategy, expansion step size, steering rate constraints, and PID parameters of the PID control algorithm. This ensures that trajectory generation and correction control are adapted to the same driver operation mode, achieving adaptive matching of trajectory planning and correction control to the driver's operation characteristics. By using Kalman filtering to predict the future steering angle and calculate the prediction deviation, the PID control algorithm is only triggered to output a corrected steering angle when the deviation exceeds a preset deviation threshold, avoiding frequent corrections that interfere with the driver's normal operation. By converting the corrected steering angle into a direction correction prompt and outputting it to the parking lot terminal 106, which then presents it to the driver, and by having the server 104 receive the micro-motion feature vector uploaded by the vehicle terminal 102 after the driver's adjustment operation as a new input to repeat the above process, a closed-loop control system for human-machine collaboration is formed. In addition, by setting the maximum number of iterations and a timeout mechanism, the algorithm's real-time response capability in various scenarios is guaranteed. The incremental computing strategy reduces the amount of redundant computing. Vehicle-road cooperation or 5G communication ensures the feasibility of the server 104 deployment scheme.

[0041] In identifying driver operating patterns, a quantifiable and repeatable judgment method is needed to ensure the objectivity and accuracy of the identification results. Traditional methods often employ fuzzy comprehensive evaluation or expert scoring to subjectively classify driver styles. These methods lack unified quantitative standards, and different implementers may arrive at different classification results, leading to a lack of reliability in subsequent adaptive adjustments. Therefore, in one embodiment, such as... Figure 3 As shown, the driver's operating pattern is identified based on the statistical characteristics of micro-motion feature vectors, including: The standard deviation of the steering wheel angle change rate and the average value of the steering delay time are calculated based on the micro-motion feature vectors at multiple consecutive time points. The standard deviation of the steering wheel angle change rate is compared with a preset standard deviation threshold, and the average steering delay time is compared with a preset average threshold. When the standard deviation of the steering wheel angle change rate is less than the standard deviation threshold and the average steering delay time is greater than the average value threshold, the driver's operation mode is identified as a conservative mode; otherwise, the driver's operation mode is identified as an aggressive mode.

[0042] Specifically, firstly, server 104 calculates the standard deviation of the steering wheel angle change rate and the average value of the steering delay time based on the micro-motion feature vectors from multiple consecutive time points. During vehicle operation, server 104 continuously receives the steering wheel angle change rate and steering delay time uploaded by vehicle terminal 102, forming a micro-motion feature vector sequence. Using the data from the most recent M sampling times (e.g., M=50, corresponding to a 5-second time window at a sampling rate of 10Hz) as an analysis window, the standard deviation of the steering wheel angle change rate data within this window is calculated, and the average value of the steering delay time data within this window is calculated. It should be noted that the specific value of M depends on the product of the sampling rate and the selected analysis time window. In this embodiment, M=50 corresponds to a 5-second window at a sampling rate of 10Hz. In practical applications, this can be adjusted according to different sampling rates, and this embodiment does not limit this.

[0043] The standard deviation of the steering wheel angle change rate can be obtained using any statistical calculation method available in the art, including but not limited to existing standard deviation calculation formulas. The standard deviation reflects the dispersion of the steering wheel angle change rate: a smaller standard deviation indicates a more stable and less variable steering wheel speed, suggesting a more stable driving style; a larger standard deviation indicates greater fluctuations and a more varied steering wheel speed, suggesting a more flexible and adaptable driving style.

[0044] The average steering delay time can be obtained by any statistical calculation method in the prior art, including but not limited to existing average calculation formulas. The average steering delay time reflects the average delay between the driver's operation and the vehicle's response: the larger the average value, the more hesitant the driver's reaction and the less decisive the operation, indicating a conservative driving style; the smaller the average value, the faster the driver's reaction and the more decisive the operation, indicating an aggressive driving style.

[0045] Then, server 104 compares the standard deviation of the calculated steering wheel angle change rate with a preset standard deviation threshold, and compares the average value of the calculated steering delay time with a preset average value threshold. The preset standard deviation threshold and preset average value threshold are empirical values ​​derived from statistical analysis of a large amount of driver operation data, used to distinguish the operating characteristics of different driving styles. These two thresholds can be set during server 104 initialization or dynamically adjusted based on actual data during use.

[0046] Finally, server 104 identifies the driver's operating mode based on the comparison results. When the standard deviation of the steering wheel angle change rate is less than a preset standard deviation threshold and the average steering delay time is greater than a preset average threshold, it indicates that the driver's operation is smooth, the steering speed change is small but the response is slow, and the driver's operating mode is identified as a conservative mode. Otherwise, when the standard deviation of the steering wheel angle change rate is greater than or equal to the preset standard deviation threshold, or the average steering delay time is less than or equal to the preset average threshold, the driver's operating mode is identified as an aggressive mode.

[0047] This embodiment quantifies the stability of the driver's steering operation by using the standard deviation of the steering wheel angle change rate and quantifies the hesitation of the driver's operation response by using the average value of the steering delay time. Based on the comparison results of these two statistics with preset thresholds, it achieves objective and quantitative identification of the driver's operation mode, avoids the uncertainty caused by subjective classification, and provides a reliable input basis for subsequent adaptive adjustment of algorithm parameters according to the operation mode.

[0048] After identifying the driver's operating mode, the parameters of the Fast Expanded Random Tree Star Algorithm need to be differentiated according to different operating modes to adapt the generated parking trajectory to different driving styles. Existing Fast Expanded Random Tree Star Algorithms typically employ fixed sampling strategies, expansion step sizes, and steering rate constraints, which cannot be adjusted according to the driver's personalized operating habits. This results in a mismatch between the generated trajectory and the driver's actual operating style, causing the driver to feel unnatural or require frequent corrections when tracking the trajectory. Therefore, in one embodiment, the sampling strategy, expansion step size, and steering rate constraints are determined based on the driver's operating mode, including: When the driver's operation mode is conservative, the sampling strategy is to randomly sample nodes are preferentially distributed in the smooth path area, the expansion step size is the first expansion step size, and the steering rate constraint is that the steering wheel angle change rate does not exceed the preset angle change rate threshold. When the driver's operating mode is the aggressive mode, the sampling strategy is that the distribution range of random sampling nodes is greater than that in the conservative mode, the expansion step size is a second expansion step size greater than the first expansion step size, and the steering rate constraint is that the steering wheel angle change rate does not exceed the sum of the angle change rate threshold and the mode coefficient.

[0049] Specifically, in the Fast Expanding Random Tree Star algorithm, the sampling strategy determines the spatial distribution of random sampling nodes in each iteration, directly affecting the directionality and efficiency of path search. When the driver's operating mode is conservative, the sampling strategy prioritizes the distribution of random sampling nodes in smooth path regions. A smooth path region refers to a spatial region where the path curvature changes little and the steering wheel angle changes relatively smoothly. In this embodiment, the smooth path region can be quantitatively determined by the curvature values ​​of each point on the path: when the absolute curvature value of each point on a path is less than a preset curvature threshold (e.g., 0.05 per meter), the area where that path segment is located is determined to be a smooth path region. The preset curvature threshold can be set according to the actual application scenario and vehicle steering performance; this embodiment does not limit this. In the actual operation of the Fast Expanding Random Tree Star algorithm, this strategy can be implemented by introducing bias during random sampling, that is, when generating random sampling nodes, the sampling points are distributed near smooth path regions with a higher probability. The expansion step size refers to the distance between the new node and the nearest node during each expansion. When the driver's operating mode is conservative, the expansion step size is set to the first expansion step size s1, which is a smaller step size value in meters (m), such as 0.3 meters to 0.5 meters. A smaller expansion step size can make the generated path more refined and smoother. The steering rate constraint ensures that the rate of change of the steering wheel angle does not exceed a preset angle change rate threshold. This threshold is a pre-set upper limit value for the rate of change of angle, in degrees per second (° / s), such as 10 degrees / second, used to limit the maximum rate of steering wheel rotation in conservative mode.

[0050] When the driver's operating mode is aggressive, the sampling strategy involves a wider distribution range of randomly sampled nodes compared to the conservative mode. In aggressive mode, the spatial distribution of randomly sampled nodes is more extensive, not limited to the vicinity of smooth path regions, allowing exploration of a larger spatial area and potentially finding shorter or more flexible paths. The expansion step size is a second expansion step size s2, greater than the first expansion step size, where s2 is in meters (m), for example, 0.8 to 1.2 meters. A larger expansion step size allows path searching to cover a larger spatial range more quickly, improving search efficiency. The steering rate constraint ensures that the rate of change of the steering wheel angle does not exceed the sum of the angle change rate threshold and the mode coefficient. The mode coefficient is an additional relaxation, in degrees per second (° / s), for example, 5 degrees / second, making the upper limit of the steering rate in aggressive mode greater than that in conservative mode, allowing for faster steering wheel operations and thus generating trajectories with more dramatic curvature changes.

[0051] It should be noted that the value of the expansion step size affects the balance between the precision and speed of the path search. The specific values ​​of the first expansion step size s1 and the second expansion step size s2 mentioned above are only examples. In practical applications, they can be reasonably set according to the vehicle size and parking space size. This application embodiment does not limit this.

[0052] This embodiment achieves adaptive matching of trajectory generation to driver's operating style by setting a smooth area bias sampling, a smaller expansion step size, and a lower steering ratio upper limit for conservative drivers. For conservative drivers, it sets a wide area sampling, a larger expansion step size, and a higher steering ratio upper limit to generate a flexible and fast trajectory that matches the decisive operating style of aggressive drivers.

[0053] After identifying the driver's operating mode, the parameters of the PID control algorithm need to be differentiated according to different operating modes to adapt the output steering angle correction to different driving styles. Traditional PID control algorithms typically use fixed proportional gain, integral gain, and derivative gain. Regardless of the driver's operating style, the output correction response characteristics remain consistent, which may make conservative drivers feel the correction is too abrupt, while aggressive drivers may feel the correction is too sluggish. Therefore, in one embodiment, the PID parameters include proportional gain, integral gain, and derivative gain; determining the PID parameters according to the driver's operating mode includes: When the driver's operating mode is conservative, the corresponding proportional gain is the first proportional gain value, the corresponding integral gain is the first integral gain value, and the corresponding derivative gain is the first derivative gain value. When the driver's operating mode is the aggressive mode, the corresponding proportional gain is the second proportional gain value which is greater than the first proportional gain value, the corresponding integral gain is the second integral gain value which is less than the first integral gain value, and the corresponding derivative gain is the second derivative gain value which is greater than the first derivative gain value.

[0054] Specifically, in the PID control algorithm, the proportional gain determines the strength of the output's response to the current error: the larger the proportional gain, the stronger the response of the steering angle correction to the current predicted deviation, and the faster the correction; the smaller the proportional gain, the smoother the response of the steering angle correction to the current predicted deviation, and the gentler the correction. The integral gain determines the degree of influence of historical error accumulation on the output: the larger the integral gain, the greater the influence of historical error accumulation on the steering angle correction, and the stronger the ability to eliminate steady-state errors, but it may lead to integral saturation; the smaller the integral gain, the smaller the influence of historical error accumulation. The derivative gain determines the degree of influence of the error change trend on the output: the larger the derivative gain, the more sensitive the response to the rate of change of deviation, and the more effectively it can suppress the rapid growth of deviation in advance, providing a feedforward correction effect.

[0055] The proportional gain is dimensionless, while the integral gain is in units of the reciprocal of a second (s). -1 The unit of differential gain is seconds (s).

[0056] When the driver's operating mode is conservative, server 104 sets the proportional gain to the first proportional gain value. The integral gain is the first integral gain value. The differential gain is the first differential gain value. For conservative drivers, a smaller proportional gain setting allows for a smoother response to steering angle deviations, preventing excessively large corrections that might feel abrupt to the driver; a moderate integral gain setting can eliminate steady-state errors without causing excessive accumulation; and a smaller derivative gain setting can prevent excessive sensitivity to the rate of change of deviations.

[0057] When the driver's operating mode is in aggressive mode, server 104 sets the proportional gain to a second proportional gain value that is greater than the first proportional gain value. (Right now The integral gain is set to a second integral gain value that is less than the first integral gain value. (Right now The differential gain is set to a second differential gain value that is greater than the first differential gain value. (Right now For aggressive drivers, a larger proportional gain setting allows for a faster and more direct response to steering angle corrections, matching the decisive driving style of aggressive drivers; a smaller integral gain setting reduces the impact of integral accumulation on rapid maneuvers, preventing excessive cumulative corrections from the integral term during frequent maneuvers; a larger derivative gain setting better responds to rapid changes in deviation, providing more timely feedforward corrections.

[0058] It should be noted that the gain parameter of the PID control algorithm affects the response speed and stability of the corrected steering angle. Its specific value can be calibrated by the Ziegler-Nichols engineering tuning method or the critical proportional method in the prior art. This application does not limit this.

[0059] This embodiment sets a smaller proportional gain and derivative gain, as well as a moderate integral gain, for conservative drivers, so that the steering angle correction response is smooth and gradual, which is in line with the gentle operating habits of conservative drivers; and sets a larger proportional gain and derivative gain, as well as a smaller integral gain, for aggressive drivers, so that the steering angle correction response is rapid and direct, which is in line with the decisive operating habits of aggressive drivers, thereby improving the acceptability and effectiveness of the steering correction prompt.

[0060] In practical applications, using only the steering wheel angle change rate and steering delay time as micro-motion features may not fully characterize the driver's operating state and the vehicle's operating environment. For example, the current vehicle speed directly affects the vehicle's motion response characteristics; the same steering wheel angle change rate produces significantly different vehicle responses at different speeds. The wheel trajectory curvature radius directly reflects the curvature of the vehicle's current trajectory and is an important parameter for assessing steering status. To enable the micro-motion feature vector to more comprehensively describe the driver's operating characteristics and vehicle operating state, in one embodiment, the micro-motion feature vector also includes the steering wheel angle, vehicle speed, and wheel trajectory curvature radius. The steering wheel angle change rate is calculated based on the collected steering wheel angle, and the wheel trajectory curvature radius is calculated based on the steering wheel angle and vehicle speed.

[0061] Specifically, steering wheel angle This refers to the rotation angle of the steering wheel relative to its center position, collected by the steering wheel angle sensor and uploaded to the server 104 via the vehicle terminal 102, with the unit being degrees (°). Vehicle speed. This refers to the vehicle's speed, which is collected by the vehicle speed sensor and uploaded to the server 104 via the vehicle terminal 102, and is measured in meters per second (m / s).

[0062] The rate of change of the steering wheel angle is calculated based on the steering wheel angle received by server 104. Specifically, the steering wheel angle value θ(t) is received at consecutive time points, and the values ​​at two adjacent time points t and t' are calculated. Dividing the steering wheel angle difference by the time interval Δt yields the rate of change of the steering wheel angle within that time interval. ,in The sampling time interval is expressed in seconds (s). The unit is degrees per second (° / s).

[0063] Wheel trajectory radius of curvature Calculated based on steering wheel angle and vehicle speed. Specifically, server 104 determines the front wheel steering angle based on steering wheel angle and steering ratio. Combined with vehicle speed and vehicle wheelbase The radius of curvature of the wheel trajectory is calculated using a vehicle kinematics model. The unit is meters (m). The radius of curvature of the wheel track reflects the curvature of the vehicle's current trajectory and is an important parameter for assessing the vehicle's steering status.

[0064] Server 104 will adjust the steering wheel angle Speed and the radius of curvature of the wheel trajectory Rate of change of steering wheel angle Steering delay time Together they form a micro-motion feature vector .

[0065] This embodiment adds steering wheel angle, vehicle speed, and wheel trajectory curvature radius to the micro-motion feature vector, enabling the micro-motion feature vector to more comprehensively describe the driver's operating characteristics (steering wheel angle and steering wheel angle change rate) and the vehicle's operating state (vehicle speed and wheel trajectory curvature radius). This provides a richer information foundation for subsequent driver pattern recognition and trajectory generation, thereby improving the accuracy of pattern recognition and the rationality of trajectory generation.

[0066] In vehicle kinematics analysis, an accurate calculation method is needed to determine the radius of curvature of the wheel trajectory as a component of the micro-motion characteristic vector. The Ackermann steering model in vehicle kinematics provides the mathematical relationship between the radius of curvature of the wheel trajectory and vehicle speed, wheelbase, and front wheel steering angle. However, in practical applications, it is necessary to clarify the acquisition methods and dimensional processing of each parameter to ensure the accuracy of the calculation results. Therefore, in one embodiment, the radius of curvature of the wheel trajectory is calculated using the following formula: in, For vehicle speed, This refers to the vehicle's wheelbase. This is the front wheel steering angle, which is determined based on the steering wheel angle and steering ratio.

[0067] Specifically, vehicle speed Data obtained by server 104 from vehicle terminal 102, in meters per second (m / s). Vehicle wheelbase. This is the distance between the front and rear axles of the vehicle, an inherent parameter of the vehicle, measured in meters (m), and can be preset during server 104 initialization. Front wheel steering angle. According to the steering wheel angle and steering ratio Determined, that is Steering ratio Steering ratio is the ratio between the steering wheel angle and the front wheel steering angle. It's an inherent parameter of the vehicle, and different types of vehicles have different steering ratios. Steering ratio and front wheel steering angle The units are all degrees (°). Substituting into the formula... When performing calculations, it is necessary to... The tangent value is calculated after converting degrees to radians. The conversion method is as follows: This formula, based on the Ackermann steering model in vehicle kinematics, describes the relationship between the radius of curvature of the wheel trajectory and vehicle speed, wheelbase, and front wheel steering angle when the vehicle is traveling at low speeds.

[0068] Server 104 can calculate the radius of curvature of the wheel trajectory at the current moment by substituting the above parameters into the formula. The unit is meters (m). The radius of curvature of the wheel track reflects the degree of curvature of the vehicle's current driving trajectory: the smaller the radius of curvature, the more severe the trajectory curvature and the greater the vehicle's steering angle; the larger the radius of curvature, the closer the trajectory is to a straight line and the smaller the vehicle's steering angle.

[0069] In practical applications, vehicle wheelbase and steering ratio It can be set according to the specific vehicle model. For example, for ordinary passenger cars, the wheelbase... The steering ratio is typically between 2.4 meters and 3.0 meters. Typically, the values ​​are between 12 and 18. It should be noted that the above ranges for wheelbase and steering ratio are merely examples; in actual applications, they should be set according to the inherent parameters of the specific vehicle model. This application embodiment does not impose any limitations on this. Furthermore, if the vehicle is already equipped with a sensor or calculation module capable of directly outputting the radius of curvature of the wheel trajectory and uploading it to the server 104 via the vehicle terminal 102, the server 104 can directly obtain this parameter without needing to calculate it using the above formula. This application embodiment does not impose any limitations on this.

[0070] In the process of generating parking trajectories using the Fast Expanding Random Tree Star algorithm, if only the geometric feasibility of the path is considered while ignoring the physical limitations of the vehicle itself, the generated trajectory may be geometrically feasible but unexecutable by the actual vehicle. For example, if the steering wheel angle required by the trajectory exceeds the maximum angle of the vehicle's steering system, or if the radius of curvature of the trajectory is too small, causing the vehicle to sideslip at the corresponding speed, the trajectory will be unexecutable in actual driving. Therefore, in one embodiment, the dynamic constraints that the node expansion of the Fast Expanding Random Tree Star algorithm must satisfy during the generation of parking trajectories also include steering wheel angle constraints and radius of curvature constraints. Steering wheel angle constraint: The absolute value of the steering wheel angle shall not exceed the maximum steering wheel angle set according to the vehicle type; Curvature radius constraint: The radius of curvature of the wheel trajectory shall not be less than... ,in, For vehicle speed, The coefficient of friction of the road surface. This is the acceleration due to gravity.

[0071] Specifically, the steering wheel angle constraint is that the absolute value of the steering wheel angle does not exceed the maximum steering wheel angle set according to the vehicle type, i.e. ,in For steering wheel angle, The values ​​represent the maximum steering wheel angle, all in degrees (°). Different types of vehicles have different maximum steering wheel angles, determined by the mechanical structure of the vehicle's steering system. For example, the maximum steering wheel angle of a typical passenger car is usually one and a half turns (approximately 540 degrees) or two turns (approximately 720 degrees). During the node expansion process of the Fast Expanding Random Tree Star algorithm, when server 104 evaluates a candidate path point, it needs to check the steering wheel angle corresponding to that point. Does it exceed the maximum steering wheel angle? If the condition is exceeded, the node is determined not to meet the constraints and will not be added to the random tree.

[0072] The radius of curvature constraint is that the radius of curvature of the wheel trajectory is not less than ,Right now The vehicle speed is measured in meters per second (m / s); the road surface friction coefficient is dimensionless; and the acceleration due to gravity is measured in meters per second squared (m / s²). 2 The speed is approximately 9.8 meters per second squared. The radius of curvature of the wheel trajectory is expressed in meters (m). This constraint is derived from the sideslip condition in vehicle dynamics. When the vehicle travels at a speed... When driving, if the radius of curvature of the trajectory is too small, the centripetal force required by the vehicle will be insufficient. It may exceed the maximum static friction force between the tire and the road surface. This caused the vehicle to skid, among which For vehicle mass. The road surface friction coefficient reflects the magnitude of the adhesion between the tire and the road surface; different types of road surfaces have different friction coefficients. The road surface friction coefficient can be obtained in any way available in the prior art. For example, server 104 can obtain the current parking lot's road surface type and its corresponding friction coefficient from the road surface status database of the parking lot management system. This road surface status database can be pre-stored in server 104's data storage system or obtained from the cloud; or it can be obtained from road status information uploaded by vehicle terminal 102 by server 104. This application embodiment does not limit the specific method of obtaining the road surface friction coefficient. During the node expansion process of the fast expanding random tree star algorithm, when server 104 evaluates a candidate path point, it needs to calculate the minimum allowable radius of curvature based on the current vehicle speed and the road surface friction coefficient. If the radius of curvature of the path point is less than the minimum permissible radius of curvature If the condition is not met, the node is determined to be non-compliant and will not be added to the random tree.

[0073] This embodiment introduces a steering wheel angle constraint. This ensures that the steering wheel angle corresponding to all path points in the generated trajectory does not exceed the mechanical limits of the vehicle's steering system; this is achieved by introducing curvature radius constraints. This ensures that the generated trajectory will not cause vehicle skidding under given vehicle speed and road conditions. The introduction of the above constraints enables the parking trajectory generated by the Fast Extended Random Tree Star Algorithm to meet the path search requirements while also satisfying the vehicle's physical limitations and driving safety requirements, thus guaranteeing the practical feasibility of the generated trajectory.

[0074] When the Fast Extended Random Tree Star Algorithm generates multiple candidate parking trajectories, an evaluation mechanism is needed to select the optimal one. Traditional methods typically use the shortest path length or the smallest curvature change as evaluation metrics, but these metrics do not consider historical successful parking experience, which may lead to a mathematically optimal trajectory that performs poorly in actual parking. Therefore, in one embodiment, the Fast Extended Random Tree Star Algorithm compares the generated parking trajectory with historical successful trajectories based on a similarity formula. Calculate similarity, where This represents the total number of time steps contained in the parking trajectory. Let i be the i-th turning angle (in degrees) of the generated parking trajectory. The i-th steering angle (in degrees) represents the standard parking posture. To select the parking trajectory with the highest similarity as the output parking trajectory based on the weight coefficients (dimensionless) determined according to the driver's operation mode.

[0075] Specifically, the Fast Expanded Random Tree Star algorithm generates multiple candidate paths from the starting point to the ending point during the search process. In each iteration, when the algorithm finds a new feasible path, server 104 compares this path with historical successful trajectories. Historical successful trajectories refer to those that successfully guided the vehicle to a standard parking posture during previous parking processes and are stored in server 104's data storage system. The purpose of the similarity comparison is to evaluate the degree of similarity between the currently generated trajectory and those that have been verified as valid.

[0076] In the similarity formula, server 104 compares the generated candidate trajectory with the standard parking trajectory point-by-point at the same time step i, calculating the absolute value of the difference between the two steering angles at each time step. All The summation of the differences at each time step is multiplied by the weighting coefficient, and then rounded to the natural constant. The negative value of an exponential function with base 0.05. Weighting coefficients. Determined based on the driver's operating mode: In conservative mode... A larger value makes the similarity score more sensitive to steering angle deviation, tending to select a path closer to the standard trajectory; in aggressive mode, A smaller value reduces the sensitivity of similarity to steering angle deviation, allowing the selection of paths that differ slightly from the standard trajectory.

[0077] Calculate the similarity Subsequently, server 104 used it as an evaluation metric for path quality. The value range is between 0 and 1: when When the value approaches 1, it indicates that the current trajectory is very similar to a historical successful trajectory; when... When the value approaches 0, it indicates that the current trajectory differs greatly from historical successful trajectories. The closer the value is to 1, the higher the path quality, and the greater the probability that the trajectory will be successfully executed. During the search process of the Fast Expanding Random Tree Star algorithm, when multiple feasible paths are generated, server 104 selects the parking trajectory with the highest similarity as the output parking trajectory.

[0078] This embodiment compares the generated trajectory with historical successful trajectories, using similarity as an evaluation index for path quality. This allows the trajectory selection process to fully utilize historical successful parking experience, prioritizing paths similar to verified effective trajectories. This improves the quality and reliability of trajectory generation and reduces the exploration time for invalid paths.

[0079] In the process of using Kalman filtering for steering angle prediction, it is necessary to clarify the specific implementation structure of Kalman filtering and the composition of the state vector. Kalman filtering is a recursive state estimation algorithm, and the specific implementation of its prediction and update steps affects the accuracy and real-time performance of the state estimation. Therefore, in one embodiment, Kalman filtering includes a prediction step and an update step. The prediction step performs one or more steps to predict the state vector based on the state transition matrix and the control input matrix. The update step corrects the prediction result based on the observation matrix and sensor measurements. The state vector includes the steering wheel angle, vehicle speed, and steering delay time.

[0080] Specifically, the prediction and update steps of the Kalman filter can be performed using any existing Kalman filter implementation, including but not limited to the existing standard Kalman filter algorithm. In the prediction step, the server 104 determines the update steps based on the state transition matrix. and control input matrix For the state vector Make predictions. State vector. It is a set of variables describing the vehicle's driving state at the k-th time step. In this embodiment, the state vector ,in The steering wheel angle at the k-th time step; Let be the vehicle speed at the k-th time step; Let be the steering delay time at the k-th time step, in seconds (s). This can be calculated using the time difference between the steering wheel angle change and the vehicle response. State transition matrix. It describes the evolution of the vehicle's driving state from the current moment to the next moment, and can be set as follows: This matrix represents: the one-step prediction of the steering wheel angle is the current angle plus the current vehicle speed multiplied by the time interval (i.e., the integral of the angle over time); the one-step prediction of the vehicle speed remains unchanged (assuming zero acceleration); and the one-step prediction of the steering delay time remains unchanged. Control Input Matrix Describes external control input The impact of (e.g., the driver's operation of the steering wheel) on the vehicle's driving state. In the prediction step, server 104 will use the current state vector... With the state transition matrix Multiply, and then add the control input matrix. With control input The product of these two values ​​yields the predicted value of the state vector at the next moment.

[0081] In the update step, server 104 adjusts the observation matrix. and sensor measurements The prediction results are corrected. Observation matrix. This describes the mapping relationship between vehicle driving status and sensor measurement values. In this embodiment, the sensor measurement values ​​received by server 104... This includes the steering wheel angle measured by the steering wheel angle sensor and the vehicle speed measured by the vehicle speed sensor, uploaded by the vehicle terminal 102, and the observation matrix. It can be set to This indicates that the steering wheel angle in the measured value directly corresponds to the first component of the state vector, and the vehicle speed directly corresponds to the second component of the state vector. In practical applications, server 104 obtains the current measured value through vehicle terminal 102. The measured value is compared with the predicted value, the difference between the two (i.e., the residual) is calculated, and then the predicted value is corrected based on the residual to obtain a more accurate state estimate.

[0082] The prediction and update steps are performed alternately, forming a recursive estimation process. At each time step, server 104 first executes the prediction step to obtain the predicted state value, and then, when a new sensor measurement value arrives, it executes the update step to correct the predicted value. Through this recursive method of prediction-update-re-prediction-re-update, Kalman filtering can track changes in the vehicle's driving state in real time and provide accurate state estimation.

[0083] This embodiment clarifies the composition of the Kalman filter's state vector (steering wheel angle, vehicle speed, and steering delay time), enabling the Kalman filter to fully capture the dynamic relationship between driver operation and vehicle response. By defining the state transition matrix, the evolution law between each state variable is clarified. By defining the observation matrix, the mapping relationship between sensor measurements and state vectors is clarified, thereby improving the accuracy of steering angle prediction.

[0084] In the process of calculating the corrected steering angle based on the predicted deviation using the PID control algorithm, it is necessary to clarify the specific calculation method for the corrected steering angle. The PID control algorithm is an error feedback-based control method, and the discretization of its proportional, integral, and derivative terms directly affects the calculation accuracy and response characteristics of the corrected steering angle. Therefore, in one embodiment, the PID control algorithm executed by the server 104 is based on the formula... Calculate the corrected steering angle, where This is the corrected steering angle at the k-th time step, in degrees (°). The prediction error at the k-th time step is expressed in degrees (°). For the proportional gain of the PID control algorithm, The integral gain of the PID control algorithm. This is the derivative gain of the PID control algorithm.

[0085] Specifically, this involves correcting the steering angle. It consists of the sum of three items.

[0086] The first item is the proportion. That is, proportional gain Multiply by the prediction bias at the current time step The proportional term's role is to generate an immediate correction based on the current deviation: the larger the deviation, the larger the correction; the smaller the deviation, the smaller the correction. The proportional term is the most direct source of correction in PID control, capable of quickly responding to changes in deviation.

[0087] The second item is an integral term. That is, integral gain Multiply by prediction bias The integral term represents the time integral from the start to the current moment. Its purpose is to accumulate historical deviations and eliminate steady-state errors: when a small deviation persists in the vehicle's driving state over a long period, the integral term gradually accumulates, generating a continuous corrective effect to ensure that the vehicle's driving state eventually converges to the desired value. However, the integral term can also lead to integral saturation, meaning that the integral amount becomes too large when the deviation persists for a long time. Therefore, it is necessary to set the integral gain appropriately.

[0088] The third term is the differential term. That is, differential gain Multiply by prediction bias The derivative with respect to time. The role of the differential term is to generate a forward-looking correction based on the changing trend of the deviation: when the deviation is increasing, the differential term generates a positive correction to suppress further increase in the deviation; when the deviation is decreasing, the differential term generates a negative correction to prevent overshoot. The differential term can respond in advance to changes in the deviation trend, improving the stability of the vehicle's driving state.

[0089] This embodiment achieves an immediate response to the current deviation through a proportional term, eliminates steady-state error through an integral term, and suppresses rapid changes in deviation and improves the stability of the vehicle's driving state through a derivative term. The combination of these three factors allows the corrected steering angle output by the PID control algorithm to both quickly respond to deviation changes and maintain stability, avoiding oscillations. By quantifying the predicted deviation into a specific corrected steering angle value, an accurate numerical basis is provided for generating direction correction prompts.

[0090] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps.

[0091] Secondly, based on the same inventive concept, this application also provides a system for implementing the above-described adaptive adjustment method for parking posture based on vehicle micro-motion feature recognition. The solution provided by this system is similar to the implementation described in the above method; therefore, specific limitations in one or more system embodiments provided below can be found in the above-described limitations of the adaptive adjustment method for parking posture based on vehicle micro-motion feature recognition, and will not be repeated here.

[0092] In one embodiment, such as Figure 4 As shown, the system includes: a micro-motion feature construction module, a trajectory planning module, a prediction and correction module, and an interactive feedback module. Among them: The micro-motion feature construction module is used to construct micro-motion feature vectors based on micro-motion features; micro-motion features include steering wheel angle change rate and steering delay time. The trajectory planning module is used to identify the driver's operating mode based on the statistical characteristics of the micro-motion feature vector at the current moment. Using the driver's operating mode as the same input, it simultaneously determines the sampling strategy, expansion step size, and steering rate constraints of the Fast Extended Random Tree Star Algorithm, as well as the PID parameters of the PID control algorithm. This ensures that the parking trajectory generated by the Fast Extended Random Tree Star Algorithm and the corrected steering angle output by the PID control algorithm are both adapted to the driver's operating mode. The Fast Extended Random Tree Star Algorithm is run with the sampling strategy, expansion step size, and steering rate constraints to generate the parking trajectory, and the expected steering angle at each time step is extracted from the parking trajectory. The prediction correction module is used to predict the steering angle for the next N time steps based on the micro-motion feature vector using Kalman filtering, and to calculate the prediction deviation between the predicted steering angle and the desired steering angle. When the absolute value of the prediction deviation exceeds the preset deviation threshold, the prediction deviation is input into the PID control algorithm. The PID control algorithm outputs the corrected steering angle with PID parameters and converts the corrected steering angle into a direction correction prompt output to the driver. The interactive feedback module collects the micro-motion feature vectors after the driver's adjustment operation, using them as new micro-motion feature vectors. This triggers the trajectory planning and prediction correction modules to execute repeatedly until the vehicle reaches a standard parking posture. Specifically, the interactive feedback module performs its own data collection function (collecting the micro-motion feature vectors after the driver's adjustment operation) in each loop.

[0093] In one embodiment, the trajectory planning module includes a pattern recognition unit. The pattern recognition unit calculates the standard deviation of the steering wheel angle change rate and the average value of the steering delay time based on micro-motion feature vectors at multiple consecutive time points; compares the standard deviation of the steering wheel angle change rate with a preset standard deviation threshold, and compares the average value of the steering delay time with a preset average value threshold; when the standard deviation of the steering wheel angle change rate is less than the standard deviation threshold and the average value of the steering delay time is greater than the average value threshold, the driver's operating mode is identified as a conservative mode; otherwise, the driver's operating mode is identified as an aggressive mode.

[0094] In one embodiment, the trajectory planning module includes an algorithm element determination unit. This unit is configured to: when the driver's operating mode is conservative, prioritize the distribution of random sampling nodes in smooth path regions, use a first expansion step size, and constrain the steering rate to ensure the steering wheel angle change rate does not exceed a preset angle change rate threshold; and when the driver's operating mode is aggressive, implement a sampling strategy where the distribution range of random sampling nodes is greater than that in the conservative mode, use a second expansion step size greater than the first expansion step size, and constrain the steering rate to ensure the steering wheel angle change rate does not exceed the sum of the angle change rate threshold and the mode coefficient.

[0095] In one embodiment, the trajectory planning module includes a control parameter determination unit. This control parameter determination unit is configured to, when the driver's operating mode is conservative, assign a first proportional gain value, a first integral gain value, and a first derivative gain value; and when the driver's operating mode is aggressive, assign a second proportional gain value greater than the first proportional gain value, a second integral gain value less than the first integral gain value, and a second derivative gain value greater than the first derivative gain value.

[0096] In one embodiment, the micro-motion feature construction module includes a steering wheel angle calculation unit, a vehicle speed acquisition unit, and a radius of curvature calculation unit. The steering wheel angle calculation unit calculates the steering wheel angle change rate based on the acquired steering wheel angle; the radius of curvature calculation unit calculates the radius of curvature of the wheel trajectory based on the steering wheel angle and vehicle speed; and the vehicle speed acquisition unit reads the vehicle speed data received from the server 104.

[0097] In one embodiment, the radius of curvature calculation unit is specifically used to calculate the radius of curvature of the wheel trajectory using the following formula: in, For vehicle speed, This refers to the vehicle's wheelbase. This is the front wheel steering angle, which is determined based on the steering wheel angle and steering ratio.

[0098] In one embodiment, the trajectory planning module includes a constraint setting unit. This constraint setting unit configures dynamic constraints that node expansion must satisfy during the generation of parking trajectories using the fast-expanding random tree-star algorithm. These constraints also include steering wheel angle constraints and radius of curvature constraints. The steering wheel angle constraint stipulates that the absolute value of the steering wheel angle does not exceed the maximum steering wheel angle set according to the vehicle type; the radius of curvature constraint stipulates that the radius of curvature of the wheel trajectory is not less than... ,in, For vehicle speed, The coefficient of friction of the road surface. This is the acceleration due to gravity.

[0099] The modules in the aforementioned parking posture adaptive adjustment system based on vehicle micro-motion feature recognition can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0100] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 5 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores all data required for system operation. The network interface communicates with external terminals via a network connection. When executed by the processor, the computer program implements a parking posture adaptive adjustment method based on vehicle micro-motion feature recognition.

[0101] Those skilled in the art will understand that Figure 5The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0102] Thirdly, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods described in the above method embodiments.

[0103] Fourthly, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the steps of any of the methods described in the above method embodiments.

[0104] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0105] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0106] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A parking posture adaptive adjustment method based on vehicle micro-motion feature recognition, characterized in that, The method includes: S1: Construct a micro-motion feature vector based on the micro-motion characteristics; the micro-motion characteristics include the steering wheel angle change rate and steering delay time; S2: Identify the driver's operating mode based on the statistical characteristics of the micro-motion feature vector at the current moment. Using the driver's operating mode as the same input, simultaneously determine the sampling strategy, expansion step size, and steering rate constraints of the Fast Extended Random Tree Star Algorithm, as well as the PID parameters of the PID control algorithm, so that the parking trajectory generated by the Fast Extended Random Tree Star Algorithm and the corrected steering angle output by the PID control algorithm are both adapted to the driver's operating mode. Run the Fast Extended Random Tree Star Algorithm with the sampling strategy, the expansion step size, and the steering rate constraints to generate a parking trajectory, and extract the expected steering angle for each time step from the parking trajectory. S3: Using Kalman filtering, predict the steering angle for the next N time steps based on the micro-motion feature vector, and calculate the prediction deviation between the predicted steering angle and the desired steering angle; when the absolute value of the prediction deviation exceeds a preset deviation threshold, input the prediction deviation into the PID control algorithm, the PID control algorithm outputs a corrected steering angle with the PID parameters, and converts the corrected steering angle into a direction correction prompt output to the driver; S4: Collect the micro-motion feature vector after the driver's adjustment operation, use it as the new micro-motion feature vector, and return to S2 until the vehicle reaches the standard parking posture.

2. The method according to claim 1, characterized in that, Identifying driver operation patterns based on the statistical characteristics of the micro-motion feature vector includes: The standard deviation of the steering wheel angle change rate and the average value of the steering delay time are calculated based on the micro-motion feature vector at multiple consecutive time points. The standard deviation of the steering wheel angle change rate is compared with a preset standard deviation threshold, and the average value of the steering delay time is compared with a preset average value threshold. When the standard deviation of the steering wheel angle change rate is less than the standard deviation threshold and the average value of the steering delay time is greater than the average value threshold, the driver operation mode is identified as a conservative mode; otherwise, the driver operation mode is identified as an aggressive mode.

3. The method according to claim 2, characterized in that, Determining the sampling strategy, the extended step size, and the steering ratio constraint based on the driver's operating mode includes: When the driver's operation mode is conservative, the sampling strategy is to randomly sample nodes preferentially distributed in the smooth path area, the expansion step size is the first expansion step size, and the steering rate constraint is that the steering wheel angle change rate does not exceed the preset angle change rate threshold. When the driver's operation mode is the aggressive mode, the sampling strategy is that the distribution range of the random sampling nodes is greater than the distribution range in the conservative mode, the expansion step size is a second expansion step size greater than the first expansion step size, and the steering rate constraint is that the steering wheel angle change rate does not exceed the sum of the angle change rate threshold and the mode coefficient.

4. The method according to claim 2, characterized in that, The PID parameters include proportional gain, integral gain, and derivative gain; the PID parameters are determined according to the driver's operating mode, including: When the driver's operating mode is the conservative mode, the corresponding proportional gain is the first proportional gain value, the corresponding integral gain is the first integral gain value, and the corresponding derivative gain is the first derivative gain value. When the driver's operating mode is the aggressive mode, the corresponding proportional gain is a second proportional gain value that is greater than the first proportional gain value, the corresponding integral gain is a second integral gain value that is less than the first integral gain value, and the corresponding derivative gain is a second derivative gain value that is greater than the first derivative gain value.

5. The method according to claim 1, characterized in that, The micro-motion feature vector also includes steering wheel angle, vehicle speed, and wheel trajectory curvature radius. The steering wheel angle change rate is calculated based on the collected steering wheel angle, and the wheel trajectory curvature radius is calculated based on the steering wheel angle and the vehicle speed.

6. The method according to claim 5, characterized in that, The radius of curvature of the wheel trajectory is calculated using the following formula: in, For the vehicle speed, This refers to the vehicle's wheelbase. The front wheel steering angle is determined based on the steering wheel angle and steering ratio.

7. The method according to claim 1 or 5, characterized in that, In the process of generating the parking trajectory, the dynamic constraints that the node expansion of the fast expanding random tree star algorithm must satisfy also include steering wheel angle constraints and radius of curvature constraints. The steering wheel angle constraint is that the absolute value of the steering wheel angle does not exceed the maximum steering wheel angle set according to the vehicle type; the radius of curvature constraint is that the radius of curvature of the wheel trajectory is not less than... ,in, For vehicle speed, The coefficient of friction of the road surface. This is the acceleration due to gravity.

8. A parking posture adaptive adjustment system based on vehicle micro-motion feature recognition, characterized in that, The system includes: The micro-motion feature construction module is used to construct a micro-motion feature vector based on micro-motion features; the micro-motion features include the steering wheel angle change rate and steering delay time. The trajectory planning module is used to identify the driver's operating mode based on the statistical characteristics of the micro-motion feature vector at the current moment. Using the driver's operating mode as the same input, it simultaneously determines the sampling strategy, expansion step size, and steering rate constraints of the Fast Extended Random Tree Star Algorithm (FRADA), as well as the PID parameters of the PID control algorithm. This ensures that the parking trajectory generated by the FRADA and the corrected steering angle output by the PID control algorithm are both adapted to the driver's operating mode. The FRADA is then run using the sampling strategy, expansion step size, and steering rate constraints to generate the parking trajectory, and the expected steering angle for each time step is extracted from the parking trajectory. The prediction correction module is used to predict the steering angle for the next N time steps based on the micro-motion feature vector using Kalman filtering, and to calculate the prediction deviation between the predicted steering angle and the desired steering angle; when the absolute value of the prediction deviation exceeds a preset deviation threshold, the prediction deviation is input into the PID control algorithm, the PID control algorithm outputs the corrected steering angle with the PID parameters, and converts the corrected steering angle into a direction correction prompt and outputs it to the driver; The interactive feedback module is used to collect the micro-motion feature vector after the driver's adjustment operation, as the new micro-motion feature vector, and trigger the trajectory planning module and the prediction correction module to be executed repeatedly until the vehicle reaches the standard parking posture.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the parking posture adaptive adjustment method based on vehicle micro-motion feature recognition as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the parking posture adaptive adjustment method based on vehicle micro-motion feature recognition as described in any one of claims 1 to 7.

Images