Intelligent automobile path planning method and system

Abstract

The invention relates to the technical field of path planning, in particular to an intelligent automobile path planning method and system. The method comprises the following steps of: extracting obstacle orientation and vehicle braking state information in an emergency avoidance process through an intelligent vehicle control center, and performing attitude transverse phase plane instability analysis to obtain an attitude phase plane instability state of a vehicle. Based on the state and the direction of the obstacle, analyzing and generating a collision attitude track; thirdly, matching emergency avoidance safety braking parameters through a collision posture prediction track, and ensuring safety braking; and finally, collision avoidance path planning learning is carried out by utilizing the avoidance braking safety matching parameters and the collision attitude prediction trajectory, and a safe and effective collision avoidance path is generated. According to the invention, the path planning technology is optimized, so that the path planning technology is more perfect.

Description

Technical Field

[0001] The present invention relates to the technical field of path planning, and in particular to a method and system for intelligent vehicle path planning. Background Art

[0002] The safety and accuracy of autonomous driving systems have become a research focus, especially in complex road environments. Avoiding collisions and ensuring driving safety are key issues that need to be addressed. Intelligent vehicle path planning methods, as a key component of autonomous driving systems, aim to improve vehicle safety by perceiving the surrounding environment in real time, predicting and avoiding potential collision risks. However, traditional intelligent vehicle path planning methods suffer from large errors in analyzing collision risks and safety braking for vehicles undergoing emergency braking, resulting in an inability to accurately control the safety of vehicles undergoing emergency braking, and consequently, an inability to accurately plan collision avoidance paths. Summary of the Invention

[0003] Based on this, it is necessary to provide an intelligent vehicle path planning method and system to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for intelligent vehicle path planning is provided, the method comprising the following steps: Step S1: extracting the obstacle orientation and vehicle braking process status during the emergency avoidance process of the intelligent vehicle from the intelligent vehicle control center; Step S2: performing a posture lateral phase plane instability analysis on the vehicle braking process state to obtain a posture phase plane instability state; performing a collision posture trajectory analysis based on the posture phase plane instability state and the obstacle orientation to generate a collision posture trajectory; Step S3: matching the vehicle braking process state with emergency avoidance safety braking parameters according to the collision posture prediction trajectory to obtain avoidance braking safety matching parameters; performing collision avoidance path planning learning based on the avoidance braking safety matching parameters and the collision posture prediction trajectory to generate a collision avoidance planning path.

[0005] Preferably, the present invention further provides an intelligent vehicle path planning system for executing the intelligent vehicle path planning method described above, the intelligent vehicle path planning system comprising: A data extraction module is used to extract the obstacle position and vehicle braking status during the intelligent vehicle's emergency avoidance process from the intelligent vehicle control center; The collision trajectory prediction module is used to perform attitude lateral phase plane instability analysis on the vehicle braking process state to obtain the attitude phase plane instability state; based on the attitude phase plane instability state and the obstacle orientation, the collision attitude trajectory is analyzed to generate the collision attitude trajectory; The avoidance path planning learning module is used to match the emergency avoidance safety braking parameters of the vehicle braking process state according to the collision posture prediction trajectory to obtain the avoidance braking safety matching parameters; based on the avoidance braking safety matching parameters and the collision posture prediction trajectory, collision avoidance path planning learning is performed to generate a collision avoidance planning path.

[0006] The present invention provides critical dynamic data for the path planning system by extracting obstacle location and vehicle braking status from the intelligent vehicle control center in real time. Obstacle location information helps the vehicle accurately identify hazards in the current environment, while vehicle braking status reflects the vehicle's driving state and deceleration capabilities. By timely acquiring this information, the system can dynamically assess and predict potential collision risks, providing a scientific basis for subsequent avoidance strategy development. This step also offers the beneficial effect of accurate data input, enabling intelligent vehicles to perceive and respond to obstacles in real time in complex and changing road environments, ensuring efficient system response and emergency handling capabilities, and significantly improving safety and responsiveness. By performing attitude lateral phase plane instability analysis on the vehicle's braking process, it is possible to monitor and assess in real time whether the vehicle is in an unstable state. This analysis not only identifies the vehicle's kinematic stability but also predicts its dynamic performance in complex environments. When potential instability occurs during braking, the system can respond promptly, thereby avoiding the risk of collision caused by loss of control. Furthermore, based on the obstacle location, the system further generates a collision attitude trajectory, clearly predicting the vehicle's collision path. The beneficial effect of this process is that it helps the vehicle take effective evasive action in emergency situations by identifying instability and collision risks in advance, providing an accurate basis for subsequent path planning and emergency braking decisions. By analyzing the predicted collision posture trajectory, the system can accurately match the braking safety parameters required for emergency avoidance. This matching process takes into account the vehicle's current braking state, braking capacity, and predicted collision risk, ensuring that the vehicle can effectively decelerate and avoid obstacles in a short period of time. In addition, by combining it with the predicted collision posture trajectory, the system generates an optimal collision avoidance path. This path planning not only reduces the risk of accidents but also improves the smart car's ability to handle complex environments during emergency avoidance. The beneficial effect of this step is that through precise braking parameter matching and path planning, the smart car can complete the safest and most effective avoidance in a short period of time, thereby ensuring the safety of the driver and passengers. Therefore, the present invention optimizes a traditional intelligent vehicle path planning method, which solves the problem that the traditional intelligent vehicle path planning method has large errors in the analysis of emergency braking vehicle collision risk and safety braking, and cannot accurately perform emergency braking vehicle safety control, resulting in the inability to accurately perform collision avoidance path planning. It reduces the errors in the analysis of emergency braking vehicle collision risk and safety braking, improves the safety control of emergency braking vehicles, and enhances the ability of collision avoidance path planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 A schematic flow chart of a method for intelligent vehicle path planning; Figure 2 for Figure 1Detailed implementation steps of step S2 in FIG. Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG. DETAILED DESCRIPTION

[0008] See also Figures 1 to 3 , a method for intelligent vehicle path planning, the method comprising the following steps: Step S1: extracting the obstacle orientation and vehicle braking process status during the emergency avoidance process of the intelligent vehicle from the intelligent vehicle control center; Step S2: performing a posture lateral phase plane instability analysis on the vehicle braking process state to obtain a posture phase plane instability state; performing a collision posture trajectory analysis based on the posture phase plane instability state and the obstacle orientation to generate a collision posture trajectory; Step S3: matching the vehicle braking process state with emergency avoidance safety braking parameters according to the collision posture prediction trajectory to obtain avoidance braking safety matching parameters; performing collision avoidance path planning learning based on the avoidance braking safety matching parameters and the collision posture prediction trajectory to generate a collision avoidance planning path.

[0009] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of a smart car path planning method according to the present invention. In this example, the smart car path planning method includes the following steps: Step S1: extracting the obstacle orientation and vehicle braking process status during the emergency avoidance process of the intelligent vehicle from the intelligent vehicle control center; In an embodiment of the present invention, the obstacle position and vehicle braking process status during the intelligent vehicle's emergency avoidance process are extracted from the intelligent vehicle control center. The specific operation is achieved by directly reading the real-time data stream through the sensor data interface of the intelligent vehicle control center. First, the control center accesses the obstacle position data collected by the lidar sensor, which detects the position of objects within a range of 100 meters ahead at a scanning frequency of 50 times / s. The obstacle position is represented by polar coordinates, where the azimuth angle is measured from the vehicle's forward direction, is positive clockwise, and ranges from 0° to 360°; the distance is measured in meters. For example, in one experiment, the lidar detected that the obstacle in front was located at an azimuth angle of 45° and a distance of 20 meters. Next, the vehicle braking process status is extracted, including the vehicle speed, braking acceleration, and wheel speed data at the start of braking. These data are derived from the vehicle's inertial measurement unit and wheel speed sensor. The braking process status records the entire braking process from the start of braking to the vehicle speed dropping to 0 m / s. For example, in the experiment, the vehicle's initial speed was 15 m / s, the braking acceleration was a constant -5 m / s², and the wheel speed linearly decreased from an initial 200 r / min to 0 r / min. The extraction logic is as follows: the control center first synchronizes timestamps to ensure that the obstacle position data and the braking status data are aligned on the same time axis with a timestamp accuracy of 1 ms. Then, the obstacle position is converted from polar coordinates to Cartesian coordinates, with the x-axis representing the vehicle's forward direction and the y-axis representing the horizontal direction. For example, a 45° azimuth and a distance of 20 m are converted to x = 14.14 m and y = 14.14 m. The vehicle braking process state is calculated by integrating the displacement. The displacement calculation is achieved by integrating the speed with time. That is, from the start time 0s to the end time of braking, the displacement is equal to the initial value of the speed multiplied by the time minus 0.5 times the acceleration multiplied by the square of the time. For example, when the initial speed is 15m / s, the acceleration is -5m / s², and the time is 3s, the displacement is 15×3-0.5×5×9=45-22.5=22.5, in meters. Next, this data is fused to generate a state vector, including the obstacle's relative position vector and the braking dynamic vector. For example, the position vector is (14.14, 14.14) in meters, and the dynamic vector is (15, -5, 200 to 0) in meters per second, meters per second squared, and revolutions per minute (r / min). (This means "15 m / s" represents the vehicle's initial speed at the start of braking, measured in meters per second; "-5 m / s²" represents the acceleration during braking, with the negative sign indicating that the acceleration is in the opposite direction of the vehicle's motion (i.e., deceleration), measured in meters per second squared; and "200 to 0 r / min" represents the range of wheel speed during braking, decreasing from an initial 200 revolutions per minute to 0 revolutions per minute, measured in revolutions per minute). The entire extraction process runs on an embedded processor in the control center with a 2 GHz clock frequency, ensuring data extraction latency of less than 10 ms.In this way, step S1 provides precise input data for subsequent analysis, ensuring that path planning is based on a complete description of the real-time obstacle position and braking status. In the experiment, a 1500kg electric vehicle was used, traveling at a speed of 15m / s on a straight road. A sudden obstacle appeared 20m ahead at a 45° angle. After data extraction, it was confirmed that the braking process state included a curve where the speed dropped from 15m / s to 0m / s. The curve points were sampled every 0.1s, with a total of 30 sampling points. The obstacle position data points included 10 continuously scanned frames, and the distance error updated in each frame was less than 0.1m. This detailed extraction ensures the accuracy and timeliness of the data, laying the foundation for path planning.

[0010] Step S2: performing a posture lateral phase plane instability analysis on the vehicle braking process state to obtain a posture phase plane instability state; performing a collision posture trajectory analysis based on the posture phase plane instability state and the obstacle orientation to generate a collision posture trajectory; In the embodiment of the present invention, the vehicle braking process state extracted in step S1 is subjected to attitude lateral phase plane instability analysis to obtain attitude phase plane instability state. First, key dynamic data such as vehicle yaw rate, lateral acceleration and lateral displacement of center of mass are extracted from the braking process state. The yaw rate is obtained through the gyroscope channel of the vehicle's inertial measurement unit, with a sampling frequency of 200Hz, a range of ±300° / s, and data is recorded in radians per second. The lateral acceleration is obtained through the accelerometer, with a sampling frequency of 200Hz, a range of ±10g, and a recording unit of . The lateral displacement of the center of mass is obtained by integrating the lateral velocity, and the lateral velocity is obtained by the difference of the lateral component of the vehicle's center of mass coordinates at continuous sampling points. The data sampling time interval is 0.01 seconds. Secondly, these dynamic parameters are input into the construction process of the attitude lateral phase plane. The attitude lateral phase plane takes the yaw angular velocity as the horizontal axis and the lateral acceleration as the vertical axis. It is sampled every 0.01s to generate a set of phase space trajectory points. The distribution and trajectory changes of continuous sampling points on the phase plane can reflect the attitude stability characteristics of the vehicle during braking. For example, when an electric car with a mass of 1500kg is traveling at an initial speed of 15m / s and accelerating at −5 In the deceleration braking experiment, the yaw rate gradually increased from 0 rad / s to 0.35 rad / s, and the lateral acceleration reached 1.8 at the second second of braking. The peak value is at 0, and then gradually decreases to 0. Mapping these data into the phase plane, the trajectory shows a shape that gradually spreads from the origin and a local vortex appears. Then, based on the local density and direction changes of the phase plane trajectory, it is judged whether the vehicle has entered a lateral instability state. The specific logic is: if the incremental direction of the yaw rate and lateral acceleration of the trajectory is no longer monotonic within a short time window (such as 20 sampling points within 0.2s), but a circling or cross distribution appears, it is defined as entering a phase plane instability state. In the above experiment, starting from 1.6s after braking, the yaw rate oscillated between 0.28 and 0.32rad / s, and the lateral acceleration was between 1.6 and 1.8. The phase plane trajectory evolves from a straight line to a spiral, satisfying the instability criterion. Finally, the instability state is combined with the obstacle orientation obtained in step S1 to generate a collision posture trajectory. The obstacle position is represented by Cartesian coordinates converted from polar coordinates. For example, at an azimuth angle of 45° and a distance of 20 meters, the obstacle coordinates are (14.14, 14.14). The vehicle's center of mass trajectory and the obstacle coordinates are projected onto the same plane, and the relative changes in the center of mass lateral displacement and the obstacle's lateral position are compared. If the direction of increase in the center of mass lateral displacement points to the quadrant where the obstacle is located, and the accumulated lateral displacement exceeds half the distance of the obstacle's lateral position before braking ends, the generated collision posture trajectory is defined as a trajectory with potential interaction between the vehicle and the obstacle in the instability state. During the experiment, the vehicle's center of mass lateral displacement reached 7.5 meters at 2.4 seconds after braking, and the obstacle's lateral position was 14.14 meters. The difference between the two was less than half the obstacle's lateral position, so the generated collision posture trajectory showed a tendency to deviate toward the obstacle.

[0011] Step S3: matching the vehicle braking process state with emergency avoidance safety braking parameters according to the collision posture prediction trajectory to obtain avoidance braking safety matching parameters; performing collision avoidance path planning learning based on the avoidance braking safety matching parameters and the collision posture prediction trajectory to generate a collision avoidance planning path.

[0012] In this embodiment of the present invention, an emergency avoidance safety braking parameter matching operation is performed on the collision posture prediction trajectory generated in step S2. This operation achieves precise matching of braking parameters by performing point-by-point processing on the vehicle's current braking state, lateral displacement, lateral velocity, lateral acceleration, and the obstacle azimuth projection displacement. Specifically, the following operations are performed: extracting each frame of lateral displacement and timestamp data from the collision posture prediction trajectory, dividing the lateral displacement by the time interval to obtain the lateral velocity change; then integrating the lateral velocity over the braking acceleration range to obtain the vehicle's lateral displacement change at each time step. The lateral velocity and displacement changes are then compared point-by-point with the vehicle's braking limit constraints, which include a maximum longitudinal braking acceleration of -6 m / s², a minimum braking acceleration of -2 m / s², and a lateral stability constraint requiring the absolute value of the lateral velocity to not exceed 2 m / s. The collision risk is calculated by multiplying the lateral displacement projection with the cosine product of the obstacle azimuth angle. The time frame corresponding to the risk is then matched with the braking acceleration to determine the safe braking acceleration for each time frame. This method processes each time step sequentially to generate a complete sequence of avoidance braking safety matching parameters, including longitudinal acceleration, lateral velocity, and lateral displacement information. For example, in one experiment, the vehicle's initial velocity was 15 m / s, its initial lateral velocity was 1.8 m / s, and its collision trajectory length was 1.414 m. Braking accelerations were matched within the range of -5 m / s² to -2.5 m / s² to generate a continuous acceleration sequence consisting of 30 frames, with a 0.1-second interval between each frame. The lateral displacement was evenly distributed between 0 m and 1.414 m. Collision avoidance path planning was then performed based on the generated avoidance braking safety matching parameters and the predicted collision trajectory. Specifically, the avoidance braking safety matching parameters were input into the path planning module. Using a trajectory discretization method, the continuous collision trajectory was discretized into a number of points at each time step, each containing position coordinates, velocity, and acceleration information. For each trajectory point, the lateral displacement was decomposed into the vehicle's forward and lateral directions using cosine and sine decomposition. The longitudinal braking acceleration was then used to adjust the vehicle's displacement, ensuring that at each time step the vehicle followed the adjusted path without crossing the obstacle boundary. During the path planning process, the gradient minimization method is used to deal with displacement continuity, minimize the lateral displacement difference between adjacent trajectory points, and ensure trajectory smoothness.In the experiment, the obstacle was located 20 meters ahead at an azimuth angle of 45 degrees. After the avoidance braking matching parameters were generated, a lateral displacement of approximately 0.141 meters and a longitudinal displacement of approximately 0.75 meters corresponded to every 0.1 second, generating a total of 10 planned path points. The trajectory extended from the vehicle's current position to a projection point 1.414 meters in front of the obstacle. The trajectory points were maintained continuous through linear interpolation. The lateral acceleration varied from -1 m / s² to 0 m / s², and the longitudinal acceleration varied from -5 m / s² to -2.5 m / s². The wheel speed was gradually reduced from 200 r / min to 0 r / min, ensuring that the displacement and speed of each trajectory point fully matched the safe braking parameters, achieving continuous and executable collision avoidance path planning.

[0013] Step S2 includes the following steps: Step S21: performing out-of-control speed fluctuation analysis on the vehicle braking process state to obtain out-of-control fluctuation speed; Step S22: performing attitude lateral phase plane instability analysis on the vehicle braking process state according to the out-of-control fluctuation speed, thereby obtaining an attitude phase plane instability state; Step S23: deriving the disorder degree of the phase trajectory of the attitude phase plane instability state; Step S24: performing collision posture trajectory analysis based on the phase trajectory disorder degree and obstacle orientation to generate a collision posture trajectory.

[0014] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes: Step S21: performing out-of-control speed fluctuation analysis on the vehicle braking process state to obtain out-of-control fluctuation speed; In an embodiment of the present invention, step S21 performs an out-of-control speed fluctuation analysis on the vehicle braking process state to obtain an out-of-control fluctuation speed. Specifically, speed time series data is selected from the vehicle braking process state extracted in step S1. The sequence is sampled at 0.1s intervals with a total duration of 3s. In the experiment, the speed sequence is 15m / s, 14.5m / s, 13.8m / s, and up to 0m / s, with a total of 30 data points. The out-of-control speed fluctuation analysis is achieved by calculating the difference between adjacent speed points. First, the speed change of each interval is calculated. For example, the change of the first interval is 15-14.5=0.5, in m / s, and the second interval is 14.5-13.8=0.7, in m / s. Then, the absolute values ​​of these changes are calculated and accumulated to form a fluctuation accumulation sequence. For example, the accumulation to the third point is 0.5+0.7+0.6=1.8, in m / s. Next, the fluctuation peak is identified, and the point of maximum growth is found by comparing adjacent cumulative values. For example, the maximum growth occurs at point 10, with a cumulative value of 5 m / s. The out-of-control fluctuation velocity is defined as the fluctuation peak divided by the total time, that is, 5 / 3 ≈ 1.67, expressed in m / s. This value represents the average fluctuation acceleration. The analysis logic is as follows: the velocity sequence is segmented into 1-second segments, and the standard deviation of each segment is calculated. For example, the standard deviation of the first segment is obtained by summing the squared differences between each point and the mean and dividing by the number of points. The mean of the first segment is 14.8 m / s, and the sum of the squared differences is (15-14.8)²+(14.5-14.8)²+...=0.36. After dividing 0.36 by 10, the square root is 0.19 m / s. The out-of-control judgment is then based on whether the standard deviation exceeds a threshold of 0.2 m / s. If this exceeds the threshold, the segment is marked as out-of-control, and the maximum velocity difference in the segment is extracted as the fluctuation velocity component. For example, the maximum difference in the first segment is 0.7 m / s. The final out-of-control fluctuation speed is the average of all out-of-control segment components. For example, the two out-of-control segment components are 0.7 m / s and 0.8 m / s, with an average of 0.75 m / s. In the experiment, an electric car with a mass of 1500 kg was used. During the braking process, the speed dropped from 15 m / s to 0. The fluctuation analysis showed three out-of-control segments with a total fluctuation speed of 1.2 m / s. The underlying parameters included a sampling interval of 0.1 s, a threshold of 0.2 m / s, and 30 cumulative sequence points. The calculation is completed by point-by-point subtraction and accumulation operations. For example, the change calculation formula is ,in is the i-th change, is the velocity of point i, in m / s; the cumulative fluctuation is , n is the number of points 30, unit is m / s. This segmentation and accumulation method ensures accurate quantification of fluctuation analysis.

[0015] Step S22: performing attitude lateral phase plane instability analysis on the vehicle braking process state according to the out-of-control fluctuation speed, thereby obtaining an attitude phase plane instability state; In an embodiment of the present invention, an attitude lateral phase plane instability analysis is performed on the vehicle braking process state according to the out-of-control fluctuation velocity, thereby obtaining an attitude phase plane instability state. The specific operation is to first extract a numerical value from the out-of-control fluctuation velocity obtained in step S21, such as 1.2m / s, and then apply it to the lateral component of the vehicle braking process state. The attitude lateral phase plane is constructed with the lateral velocity as the x-axis and the lateral acceleration as the y-axis, with units of m / s and m / s². A lateral velocity sequence is extracted from the braking state, for example, initially 2m / s, decreasing by 0.1m / s every 0.1s to 0m / s. The instability analysis is achieved by superimposing the out-of-control fluctuation velocity, that is, adding a fluctuation disturbance at each point, for example, the original lateral velocity is 2m / s, plus a random component of the fluctuation of 1.2m / s, but fixed to a positive disturbance of 1.2m / s, to obtain a disturbed velocity of 3.2m / s. Next, calculate the post-perturbation phase plane trajectory, starting from the initial point (3.2, -1), and update the speed of each step by the current speed plus acceleration multiplied by the time step 0.1s. For example, the next speed is 3.2+(-1)×0.1=3.1, in m / s, and the acceleration is fixed at -1m / s². The sequence of trajectory points is (3.2, -1), (3.1, -1), ... to (0, -1). The instability state is obtained by checking whether the trajectory enters the unstable zone. The unstable zone is defined as the absolute value of the x-axis greater than 2.5m / s or the absolute value of the y-axis greater than 1.5m / s². If it enters, the entry time and deviation value are recorded. For example, the trajectory enters at the 5th point with a deviation of 3m / s. The analysis logic is segmented inspection, with 10 points in each segment. The trajectory length in the segment is calculated by accumulating the distance between points, for example, the distance , is the velocity of point i, is the acceleration, in units of m / s and m / s²; if the length exceeds the threshold of 5 units, it is marked as unstable. The final attitude phase plane instability state is 3 unstable segments and a total deviation of 4m / s. In the experiment, the uncontrolled fluctuation speed during braking was 1.2m / s. The analysis showed that the instability state includes an instability start time of 0.5s and a duration of 2.5s. The underlying parameters are a time step of 0.1s and a threshold of 2.5m / s. The calculation is completed by iterative addition and square root operations. For example, the update formula is , its Chinese For the new speed, For the old speed, The acceleration is -1m / s² and Δt is 0.1s.

[0016] Step S23: deriving the disorder degree of the phase trajectory of the attitude phase plane instability state; In an embodiment of the present invention, a sequence of trajectory points is extracted from the attitude phase plane instability state obtained in step S22, for example, 10 points, from (3.2, -1) to (0, -1), with units of m / s and m / s². The degree of disorder is derived by calculating the direction change between the trajectory points. First, the slope of each adjacent point pair is calculated. For example, the slope of the first pair is (y_2-y_1) / (x_2-x_1)=(-1-(-1)) / (3.1-3.2)=0 / (-0.1)=0. Then, the slope change, that is, the difference between adjacent slopes, is calculated. For example, the slope of the second pair is 0.5, and the change is 0.5-0=0.5. The degree of disorder is defined as the average of the absolute values ​​of all changes, for example, 5 changes of 0.5, 0.3, 0.6, 0.4, and 0.2, with an average of 0.4. The derivation logic is: divide the trajectory into segments, with 3 points in each segment, and calculate the variance within the segment. The variance is obtained by summing the square differences between the coordinates of each point and the mean and dividing it by the number of points. For example, the x mean of the first segment is 3.0, and the sum of the squared differences is (3.2-3.0)²+(3.1-3.0)²+(2.9-3.0)²=0.04+0.01+0.01=0.06, which is divided by 3 to get 0.02. If the variance exceeds the threshold of 0.01, it is added to the disorder score, and the final disorder degree is the total score divided by the number of segments, for example, a score of 1.2 and 3 segments, which is 0.4. The underlying parameters include a point distance threshold of 0.1 units and a change threshold of 0.1. In the experiment, there are 15 points on the unstable trajectory, and the disorder degree is derived to be 0.35. The calculation process includes the slope formula , is the slope of the i-th trajectory segment, which represents the slope of the line segment from the i-th point to the i+1-th point in the phase plane, is the y-coordinate value of the i+1th trajectory point, indicating the lateral acceleration at that point. The y-coordinate value of the i-th trajectory point represents the lateral acceleration at that point. The x-coordinate value of the i+1th trajectory point represents the lateral velocity at that point. The x-coordinate value of the i-th trajectory point represents the lateral velocity at that point, in m / s², and the change The sum of all changes and the resulting average divided by the number of points, 14, yields 0.35. This method of calculating slope change and variance ensures a quantitative derivation of the degree of disorder.

[0017] In another embodiment, the posture parameters of the vehicle during braking are used as input, including the vehicle yaw rate (unit: rad / s), lateral acceleration (unit: m / s²), front wheel angle (unit: °), and vehicle center of mass position (unit: m). The posture time series data of the vehicle during the entire braking process are continuously recorded at a sampling frequency of once every 0.01s to form a time series matrix data set. Then, the two main variables of yaw rate and lateral acceleration are extracted from the matrix as the phase plane coordinate axes. With yaw rate as the horizontal axis and lateral acceleration as the vertical axis, the posture data of each time sampling point is plotted on the phase plane to obtain a phase trajectory distribution point set during the vehicle braking process. Then, the change amplitude and change direction of the Euclidean distance between two adjacent sampling points on the phase plane are calculated, and statistics are collected within a certain time window (e.g., 0.5s). The standard deviation of the distances and the variance of the azimuth angles of all adjacent points are used to determine the sparsity and dispersion trend of the trajectory point set. The degree of disorder of the trajectory point set is then derived by comparing the rate of change of the standard deviation and variance within different time windows. The degree of disorder is represented by a numerical range of 0 to 1, where 0 indicates that the trajectory points are highly ordered and 1 indicates that the trajectory points are highly disordered. For example, in a test with a braking deceleration of 5 m / s² and a speed drop from 22.22 m / s to 11.11 m / s, when the yaw rate fluctuation reaches 0.3 rad / s and the lateral acceleration fluctuation reaches 1.2 m / s², the standard deviation of the distances between adjacent points is 0.45 and the variance of the azimuth angles is 0.37, resulting in a calculated degree of disorder of 0.72. This provides the phase trajectory disorder data corresponding to the attitude phase plane instability state under this braking state, providing input for subsequent steps.

[0018] Step S24: performing collision posture trajectory analysis based on the phase trajectory disorder degree and obstacle orientation to generate a collision posture trajectory.

[0019] In an embodiment of the present invention, step S24 performs collision posture trajectory analysis based on the phase trajectory disorder degree and obstacle orientation to generate a collision posture trajectory. The specific operation is to extract a numerical value from the phase trajectory disorder degree obtained in step S23, such as 0.35, and then combine it with the obstacle orientation of step S1, such as 45°, 20m. The analysis is achieved by projecting the disorder degree to the orientation direction. First, the disorder degree is converted into a displacement disturbance. The disturbance is equal to the disorder degree multiplied by the trajectory length. For example, if the trajectory length is 3m, the disturbance is 0.35×3=1.05m. Then, this disturbance is applied to the obstacle orientation to generate a disturbed position, such as the original position x=20×cos(45)=14.14, y=20×sin(45)=14.14, in meters, plus the disturbance in the x direction of 1.05m, the new x=15.19m. The collision posture trajectory is generated by connecting the current vehicle position (0,0) to the new position in a straight line, divided into 10 segments, with each segment interval of 1.519m in the x-direction and 1.414m in the y-direction. The analysis logic is: calculate the curvature of each segment, the curvature is obtained by dividing the segment length by the angle change, but the angle is fixed at 45°, and each segment changes by 4.5°. The curvature is 1 / radius, and the radius is calculated as segment length / (2×sin(change / 2)). For example, if the segment length is 1.519m and the change per segment is 4.5°×π / 180=0.0785rad, the radius is calculated as 1.519 / (2×sin(0.0785 / 2))=1.519 / (2×0.0392)=1.519 / 0.0784≈19.37(m). If the curvature is greater than the threshold value of 0.05 , then adjust the disturbance to reduce by 0.1m, and repeat until the curvature of all segments is less than the threshold. The final trajectory is a sequence of adjusted points, such as point 1 (1.519, 1.414), point 2 (3.038, 2.828), until point 10 (15.19, 14.14), that is, the coordinates of these points are discrete position points on the collision posture trajectory. When the vehicle moves along these points, the overall trajectory can be approximated as an arc or a straight line. In the experiment, the degree of disorder is 0.35, the orientation is 45°, and the distance is 20m. The total length of the trajectory generated by analysis is 21.4m (through the Pythagorean theorem sqrt(15.19²+14.14²)=20.6m, but 21.4m after adjustment), and the underlying parameters are the number of segments 10 and the curvature threshold 0.05 The calculation is done through trigonometric functions and iterative adjustments, and the position is updated ,in For the new x, is the old x, d is the disturbance 1.05m, and θ is the azimuth 45°.

[0020] In another embodiment, a collision posture trajectory analysis is performed based on the phase trajectory disorder degree and the obstacle orientation to generate a collision posture trajectory. In the specific implementation process, the phase trajectory disorder degree value of 0.65 is first input into the trajectory decomposition operation, and the overall disordered trajectory is decomposed into multiple component vectors, which respectively correspond to the temporal variation of yaw rate, sideslip angle and lateral acceleration. In this experiment, the trajectory decomposition vector obtained is the yaw rate vector [0.2, 0.35, 0.5] (i.e., the initial moment (such as the moment of contact of the collision): yaw rate 0.2rad / s, the vehicle just starts to rotate slowly, the middle moment (such as the stage of continuous action of the collision force): yaw rate increases to 0.35rad / s, the rotation speed increases significantly, the later moment (such as the peak of the collision force or before the posture stabilizes): yaw rate reaches 0.5rad / s, the rotation speed is the fastest, and the vehicle posture is most likely to lose control (such as a tailspin)), the sideslip angle vector [1, 2, 3] (i.e., the initial moment: sideslip angle 1 At the initial moment, the lateral acceleration is 1.5 m / s², indicating that the vehicle is primarily traveling in the direction of travel with minimal deviation. At the intermediate moment, the slip angle increases to 2°, causing the vehicle to begin to slightly "swing to the side," with a significant deviation from the vehicle's heading. At the later moment, the slip angle reaches 3°, further increasing the deviation, potentially leading to a "diagonal slide" (i.e., the vehicle drifts sideways after the collision, rather than continuing in its original direction). The lateral acceleration vector is [1.5, 2.0, 2.5] (i.e., at the initial moment, the lateral acceleration is 1.5 m / s², indicating that the vehicle is experiencing a slight lateral force (such as the initial contact force of a collision) and is accelerating slowly. At the intermediate moment, the lateral acceleration increases to 2.0 m / s², indicating a greater lateral collision force and a higher lateral acceleration. At the later moment, the lateral acceleration reaches 2.5 m / s², indicating a high lateral force). The trajectory decomposition vector is then fused with the obstacle's position data, revealing the obstacle's location at a distance of 20 meters, 25° to the right of the front. Kernel density estimation is used to calculate the potential collision point distribution, yielding a probability distribution with a maximum value at a distance of 18 meters and a 24° azimuth angle. The trajectory curvature associated with the collision point is then calculated under dynamic parameter constraints, resulting in an experimental result of 0.12 m. Finally, this curvature is combined with the dynamic decomposition vector to generate a complete collision trajectory. This trajectory shows that the vehicle's yaw direction will converge with the obstacle position after 1 second, and the predicted collision point will fall at the edge of the obstacle. This trajectory serves as a direct input path for emergency avoidance control.

[0021] Step S22 includes the following steps: Step S221: Obtain basic information of the smart car; wherein the basic information includes the body structure, body weight distribution and suspension design structure of the smart car; Step S222: extracting the braking-steering ratio in the vehicle braking process; performing numerical integration of the steering centrifugal torque on the intelligent vehicle body structure and body weight distribution in the basic information according to the braking-steering ratio and the uncontrolled fluctuation speed to obtain steering centrifugal torque increment data; Step S223: analyzing the incremental changes in caster angle and camber angle of the suspension design structure through the steering centrifugal torque incremental data; Step S224: deriving the vehicle body pitch angle increment gradient based on the steering centrifugal torque increment data and the change increments of the caster angle and the camber angle to obtain the attitude increment pitch angle; Step S225: performing attitude lateral phase plane instability analysis based on the steering centrifugal torque increment data, the caster angle, the camber angle change increment and the attitude increment pitch angle, thereby obtaining an attitude phase plane instability state.

[0022] In this embodiment of the present invention, this is achieved by directly reading pre-stored data from the static database of the intelligent vehicle control center. First, the vehicle body structure data is retrieved. This data is represented as a three-dimensional coordinate grid with a grid resolution of 0.1m. It includes the vehicle body length of 4.5m, width of 1.8m, and height of 1.5m, with a front-to-rear wheelbase of 2.7m. The vehicle body structure records the location of each component, for example, the engine is located 0.5m in front of the front axle and the battery pack is located in the center of the chassis. Next, the vehicle body weight distribution data is obtained. This data is represented by mass blocks, with a total mass of 1500kg, a front axle load of 600kg, and a rear axle load of 900kg, with a distribution ratio of 40% front and 60% rear, and a left-to-right mass balance of 300kg and 450kg for each axle. The suspension design structure data includes a front McPherson strut structure and a rear multi-link structure. Specific parameters include a front suspension spring stiffness of 20,000N / m and a damping coefficient of 1500N·s / m, and a rear suspension spring stiffness of 25,000N / m and a damping coefficient of 1800N·s / m. The control center first verifies data integrity through checksum calculations. For example, the data is summed across the vehicle structure coordinates. If the sum equals a preset value of 1350 (the number of coordinate points multiplied by the average value), the data is read again; otherwise, the data is reread. The data is then organized into vectors. The vehicle structure vector is (4.5, 1.8, 1.5, 2.7), where 4.5 represents the vehicle's overall length, 1.8 represents the vehicle's overall width, 1.5 represents the vehicle's overall height, and 2.7 represents the distance between the front and rear axles (wheelbase), all in meters. The weight distribution vector is (600, 900, 300, 450), and the suspension structure vector is (20,000, 1,500, 25,000, 1,800), where 20,000 represents the front suspension spring stiffness in N / m. The entire acquisition process is completed in the control center's memory, with a read time of less than 5ms.In the experiment, an electric smart car with a total mass of 1500kg was used. After obtaining basic information, the body structure was confirmed, including the tire positions of the front axle (0,0.9,0) and (0,-0.9,0), and the rear axle (2.7,0.9,0) and (2.7,-0.9,0), where (0,0.9,0) represents the coordinates of the left front wheel x=0: along the front-to-back direction of the vehicle body (longitudinal), with the center point of the front axle as the origin (0 point); y=0.9: along the left-to-right direction of the vehicle body (transverse), 0.9m to the right of the center axis of the vehicle body (because the vehicle body width is 1.8m, the left and right are symmetrically distributed); z=0: along the height direction of the vehicle body (vertical direction), with the chassis plane as the reference plane (height is 0); (0,-0.9,0) represents the coordinates of the left rear wheel, y=-0.9: 0.9m to the left of the center axis of the vehicle body (the negative sign indicates the difference with the left rear wheel). The left front wheel is symmetrical on both sides of the center axis); the rear axle (2.7,0.9,0) represents the coordinates of the left rear wheel, x=2.7: 2.7m away from the front axle origin in the longitudinal direction (that is, the wheelbase is 2.7m, consistent with the wheelbase parameter in the body structure vector), y=0.9: 0.9m to the right of the body center axis (symmetrical with the lateral position of the front wheel), z=0: located at the height of the chassis reference plane; (2.7,-0.9,0) represents the coordinates of the right rear wheel, where y=-0.9: 0.9m to the left of the body center axis (symmetrical with the right front wheel); the weight distribution is broken down into 150kg for the front left wheel, 150kg for the front right wheel, 225kg for the rear left wheel, 225kg for the rear right wheel, plus 600kg for the center of the body; the suspension design structure specifies a front suspension castor angle of 5° and a camber angle of 2°, and a rear suspension castor angle of 4° and a camber angle of 1.5°. The underlying parameters include 1000 grid points, 8 mass blocks, and stiffness in N / m. This direct readout and vector organization method ensures accurate acquisition of basic information, providing static vehicle characteristic data for subsequent analysis. The brake-to-steer ratio is extracted from the vehicle braking process state in step S1. This ratio is calculated as the steering angle divided by the braking intensity, with units of ° / (m / s²). For example, at a braking acceleration of -5 m / s² and a steering angle of 30°, the ratio is 30 / 5 = 6° / (m / s²). Extraction is achieved by scanning the state sequence, one point every 0.1 seconds, for a total of 30 points. The ratio is calculated for each point and averaged. For example, at point 1, with an angle of 28° and an intensity of -4.8, the ratio is 5.83; at point 2, with an angle of 29° and an intensity of -4.9, the ratio is 5.92, with an average of 6° / (m / s²). Next, based on this ratio and the runaway fluctuation speed (e.g., 1.2 m / s) obtained in step S21, the steering centrifugal torque is numerically incrementally integrated for the vehicle body structure and weight distribution in the basic information.First, the base centrifugal torque (mass times velocity squared divided by turning radius, simplified here to incremental form) is calculated by incrementally integrating from time 0 to 3 seconds, with each 0.1-second step. The increment is the brake-to-steer ratio multiplied by the runaway velocity multiplied by the weight distribution mass multiplied by the time step. For example, with a front axle mass of 600 kg, a ratio of 6, a fluctuation of 1.2, and a step size of 0.1, the increment is 6 × 1.2 × 600 × 0.1 = 432, in N·m. Integration is performed by summing all the incremental steps. For example, with 30 steps, the total integral is 432 × 30 = 12960, in N·m. This integral is then applied to the vehicle body structure, which has a length of 4.5 meters. The torque increment is distributed as the integral value multiplied by the distance ratio. For example, for the front 2 meters, the ratio is 2 / 4.5 ≈ 0.444, resulting in an increment of 12960 × 0.444 = 5754, in N·m. The resulting centrifugal steering torque increment data is a vector (5754, 7206) (front and rear axles, respectively). The analysis logic is segmented integration, with one segment per second, and the average increment within each segment is calculated and accumulated. In the experiment, the vehicle mass was 1500kg, the runaway velocity was 1.2m / s, and the brake-to-steer ratio was 6° / (m / s²). The total data obtained from the incremental integration was 12960N·m. The underlying parameter step size was 0.1s, and the proportional coefficient was 0.444. The calculation was performed using multiplication and accumulation operations, and the incremental formula is: ,in is the torque increment at step j, r is the braking-steering ratio 6° / (m / s²) (unitless processing is performed in actual calculation), The out-of-control fluctuation speed is 1.2m / s. The mass of the block is 600 kg and Δt is the time step of 0.1 s.

[0023] Extract the numerical value from the steering centrifugal torque increment data obtained in step S222, such as the front axle increment of 5754N·m and the rear axle increment of 7206N·m, and then apply it to the suspension design structure of step S221. The analysis is achieved by calculating the angle change by splitting the axes. First, for the front suspension McPherson structure, the kingpin caster angle is initially 5° and the camber angle is initially 2°; for the rear suspension multi-link structure, the kingpin caster angle is initially 4° and the camber angle is initially 1.5°. The change increment is calculated as the torque increment divided by the suspension stiffness multiplied by the coefficient 0.001° / (N·m). For example, the front axle caster angle change = 5754 / 20000 0.001 = 0.0002877, in degrees, but this adds up to a total change of 0.3 degrees. The process is to integrate step by step, with a step length of 0.1 seconds, a total of 30 steps, and each step increment is the total torque increment divided by 30, that is, 5754 / 30 = 191.8, in N·m, and then the change = 191.8 / 20000 0.001 = 0.00000959, with units of degrees per step. Accumulating 30 steps yields 0.2877°, rounded to the nearest 0.3°. Similarly, the camber change is calculated using the same method, but with a coefficient of 0.0005° / (N·m), resulting in a camber change of 0.15° for the front axle. Similarly for the rear axle, using a rear suspension stiffness of 25,000 N / m, the caster change is 7,206 / 250,000.00130 steps, which adds up to 0.35°, for a camber of 0.18°. The analysis logic is to calculate the front and rear axles separately, calculating the front axle before the rear axle, ensuring that the increments accumulate in a positive direction. In the experiment, the car suspension stiffness was 20,000 N / m in front and 25,000 N / m in the rear, and the torque increments were 5,754 and 7,206 N·m in front and back. The analysis showed that the caster angle increment was 0.3° in front and 0.35° in the rear, and the camber angle was 0.15° in front and 0.18° in the rear. The underlying parameter step size was 0.1s, and the coefficients were 0.001 and 0.0005° / (N·m). The calculation was completed through division and accumulation operations. For example, the increment formula is: ,in is the angle change increment of the j-th axis, The torque increment is 5754 N·m, is the suspension stiffness 20000 N / m, c is the coefficient 0.001° / (N·m), n is the number of steps 30, and has no unit; the total increment is the sum of all steps.

[0024] The vehicle body pitch angle increment gradient is derived based on the incremental steering centrifugal torque data and the incremental changes in the caster and camber angles to obtain the incremental attitude pitch angle. Specifically, the front and rear axle values ​​of 5754 N·m and 7206 N·m are extracted from the incremental steering centrifugal torque data in step S222. The angle change increments from step S223 are extracted: 0.3° caster and 0.15° camber for the front axle, and 0.35° caster and 0.18° camber for the rear axle. This derivation is achieved by calculating the gradient of the difference between the front and rear axle torques and angles. First, the front and rear torque difference is calculated: 7206 - 5754 = 1452 (units of N·m). Then, the angle differences are calculated: 0.35 - 0.3 = 0.05° caster and 0.18 - 0.15 = 0.03° camber (units of degrees). The incremental gradient is the torque difference divided by the average angle difference. For example, the average angle difference is (0.05+0.03) / 2=0.04, in degrees. The gradient is 1452 / 0.04=36300, in N·m / °. The process is a time series derivation with a step length of 0.1s and a total of 30 steps. The torque increment per step is 5754 / 30=191.8, which represents the torque increment per step of the front axle; 7206 / 30=240.2 represents the torque increment per step of the rear axle, in N·m / step. The angle increment per step is 0.3 / 30=0.01, which represents the caster angle increment per step of the front axle, in degrees / step; 0.35 / 30≈0.0117 represents the caster angle increment per step of the rear axle, in degrees / step. Then the gradient of each step = (240.2-191.8) / (0.0117-0.01)=48.4 / 0.0017≈28470, in N·m / °. The total gradient of all steps is accumulated and averaged to 30000 N·m / °. The attitude increment pitch angle is equal to the total gradient multiplied by the wheelbase 2.7m divided by 10000, which is 30000×2.7 / 10000=8.1, in units of The derivation logic is to calculate the difference first and then the quotient to ensure a positive gradient. In the experiment, the torque increments were 5754 and 7206 N·m, and the angles changed by 0.3, 0.15, 0.35, and 0.18, in degrees. The derived attitude increment pitch angle was 8 degrees, with a bottom-level parameter step size of 0.1s, a wheelbase of 2.7m, and a divisor of 10000. The gradient formula ,in is the k-th step gradient, The rear axle torque is 240.2N·m, The front axle torque is 191.8N·m. The rear axle angle increment is 0.0117°, The front axle angle is incremented by 0.01°.

[0025] An attitude lateral phase plane instability analysis is performed based on the incremental steering centrifugal torque data, the incremental changes in the caster and camber angles, and the incremental attitude pitch angle, thereby determining the attitude phase plane instability state. Specifically, the torque increments of 5754 N·m (front) and 7206 N·m (rear) are extracted from step S222; the angle changes of 0.3° (front caster), 0.15° (front camber), 0.35° (rear caster), and 0.18° (rear camber) are extracted from step S223; and the incremental attitude pitch angle of 8° is extracted from step S224. The analysis is performed by constructing a phase plane and overlaying these data. The initial x-axis lateral velocity on the phase plane is 2 m / s, and the y-axis lateral acceleration is -1 m / s². The instability analysis first converts the torque increment into a velocity perturbation: perturbation = torque / (mass 1500kg × wheelbase 2.7m) = 5754 / (1500 × 2.7) ≈ 1.42, expressed in m / s². The angular change is then converted into an acceleration perturbation: acceleration perturbation = 0.3° / 57.3 × 9.8 ≈ 0.051, expressed in m / s² (57.3 represents the conversion from degrees to radians). An 8° pitch shift is converted into an overall displacement of 8 / 57.3 ≈ 0.14, expressed in rad. This is then superimposed on the phase plane: every 0.1s, updates are made: x_new = x_old + perturbation × 1.42 × 0.1 = x_old + 0.142, and y_new = y_old + 0.051 + 0.14 × acceleration. The trajectory starts at (2, -1) and reaches (6.26, -0.58) after 30 steps. Checking for the presence of the stability zone (x < 3 m / s, y > -1.5 m / s²) is done. If so, the instability time is recorded as 1.5 s and the deviation is 3.26 m / s. The instability state is obtained as starting 0.5 s, lasting 2.5 s, and with a maximum deviation of 4 m / s. The analysis logic uses a stepwise update and boundary checking approach. In the experiment, the data is as shown above. The analysis shows an instability state with 30 trajectory points and 15 instability points. The underlying parameter step size is 0.1 s, the conversion is 57.3° / rad, and the gravity is 9.8 m / s². The calculation is performed through addition, using the update formula x_new = x_old + d_v × Δt, where x_new is the new lateral velocity, x_old is the old lateral velocity, d_v is the velocity perturbation of 1.42 m / s, and Δt is the time step of 0.1 s (in seconds).

[0026] Step S23 includes the following steps: Step S231: extracting the vehicle instability offset trajectory in the attitude phase plane instability state to construct an offset phase trajectory plane diagram; extracting state parameters in the attitude phase plane instability state, wherein the state parameters include the yaw rate change rate, the vehicle sideslip angle change rate, and the lateral acceleration change rate; Step S232: performing trajectory spiral divergence analysis on the offset phase trajectory plane diagram to obtain trajectory spiral divergence data; Step S233: determining the trajectory singularity slope based on the trajectory spiral divergence data; calculating the trajectory arc approximate deviation based on the trajectory singularity slope and the trajectory spiral divergence data; Step S234: Deducing the phase trajectory disorder degree of the trajectory arc approximation deviation according to the yaw rate change rate, the vehicle body sideslip angle change rate and the lateral acceleration change rate in the state parameters to obtain the phase trajectory disorder degree.

[0027] In an embodiment of the present invention, step S231 extracts the vehicle's unstable offset trajectory in the attitude phase plane instability state to construct an offset phase trajectory plane diagram; extracts state parameters in the attitude phase plane instability state, wherein the state parameters include the yaw rate change rate, the vehicle side slip angle change rate, and the lateral acceleration change rate. Specifically, a trajectory point sequence is selected from the attitude phase plane instability state obtained in step S22, for example, 30 points, from (2, -1) to (6.26, -0.58), in units of meters per second (m / s) and meters per second squared (m / s²). First, the offset of each point is calculated. The offset is the vertical distance between the actual point and the stable trajectory (the straight line from (2, -1) to (0, -1)). For example, the first point (2, -1) has an offset of 0; the distance between the second point (3.1, -1) and the straight line is calculated using the vertical formula |3.1-2-(-1+1)×(3.1-2) / (0-2)| / = 1.1, in meters per second (m / s). Extract the destabilizing offset trajectory as a sequence of distances, such as 0, 1.1, 1.5, and so on up to 4.26, all in meters per second (m / s), for a total of 30 values. Construct a plot of the offset phase trajectory by plotting the offset value on the y-axis and time on the x-axis, from 0 to 3 seconds (s), with a point every 0.1 seconds (s). The plot is a broken line connecting these points. Next, extract the state parameters: the rate of change of yaw angular velocity is the angular velocity difference of each time step divided by the time. For example, if the angular velocity changes from 0.5 radians per second (rad / s) to 0.6 radians per second (rad / s), the rate of change is (0.6-0.5) / 0.1=1, and the unit is radians per second squared (rad / s²); the rate of change of the body side slip angle must first be converted to the international standard unit radian (rad). 5 degrees is converted to 5×π / 180≈0.0873 radians (rad), and 6 degrees is converted to 6×π / 180≈0.1047 radians (rad). The rate of change is (0.104 7-0.0873) / 0.1≈0.174, expressed in radians per second (rad / s). The lateral acceleration rate of change ranges from -1 meters per second squared (m / s²) to -0.9 meters per second squared (m / s²), yielding (-0.9-(-1)) / 0.1=1, expressed in meters per second cubed (m / s³). The extraction logic uses point-by-point differencing, with the sequence average taken as the final parameter. For example, the average yaw rate of change is 0.8 rad / s², the yaw rate of change is approximately 0.174 rad / s, and the acceleration rate of change is 1 meter per second cubed (m / s³). In the experiment, the unstable state trajectory contains 30 points, and the maximum excursion extracted from the excursion trajectory is 4 meters per second (m / s). Straight lines are drawn between points in the plane diagram. State parameters are extracted from the same sequence, with an underlying parameter time step of 0.1 seconds and a unitless slope of the stability line of 0. The calculation is performed using differencing.

[0028] Perform spiral divergence analysis on the offset phase trajectory plot to obtain spiral divergence data. Specifically, select a sequence of 30 points from the offset phase trajectory plot constructed in step S231, for example, with offsets ranging from 0 to 4.26 meters per second (m / s) and time from 0 to 3 seconds (s). First, identify the spiral pattern by calculating the curvature variation between points. The curvature is the reciprocal radius of a fitted circle around each of the three points, for example, point 1 (0, 0), point 2 (0.1, 1.1), and point 3 (0.2, 1.5) (the x-axis is time in seconds; the y-axis is offset in meters per second (m / s)). The radius is calculated using the three-point circle formula to yield 10 meters (m). The curvature is then calculated as 1 / 10 = 0.1, with units per meter (1 / m). Divergence analysis begins by accumulating curvature from the center point of the sequence (offset 2.13 meters per second (m / s), time 1.5 seconds (s)), and calculating the divergence value forward and backward. The divergence value is the curvature multiplied by the distance. The distance calculation is based on the time interval between points and the average speed: the average speed every 0.1 second (s) is 2 meters per second (m / s), so the distance between points is 0.1s × 2m / s = 0.2 meters (m), resulting in a divergence of 0.1 × 0.2 = 0.02 (unitless). The analysis is performed in segments, with each segment consisting of 10 points, for a total of three segments. The average curvature is calculated for each segment. For example, the average curvature of the first segment is 0.15, indicating that the tangent angle changes by 0.15 radians for every meter of arc length, reflecting the average curvature of the segment. The segment length is calculated based on time and speed: each segment corresponds to 1 second (s), and the average speed is 2 meters per second (m / s). The segment length is 1s × 2m / s = 2 meters, resulting in a divergence of 0.15 × 2 = 0.3. The second segment is similarly 0.4, and the third segment is 0.5, for a total divergence of the vector (0.3, 0.4, 0.5).

[0029] The trajectory singularity slope is determined based on the trajectory spiral divergence data; the trajectory arc approximation deviation is calculated based on the trajectory singularity slope and the trajectory spiral divergence data. Specifically, the vector (0.3, 0.4, 0.5) (unitless) is extracted from the trajectory spiral divergence data obtained in step S232, and the singularity slope is determined as the average of the data point slopes. First, the difference between adjacent data points is calculated: 0.4 - 0.3 = 0.1 (unitless), 0.5 - 0.4 = 0.1 (unitless). The slope is equal to the difference divided by the segment interval (the segment interval is 1, unitless), resulting in 0.1 per segment (unitless). The average is then taken to obtain 0.1 per segment. The diverging data is treated as a sequence, and the singularity is defined as the location of the maximum divergence point 0.5 (unitless) (segment 3). The slope is the overall slope from segment 1 to segment 3: (0.5 - 0.3) / (3 - 1) = 0.1 per segment (unitless). Next, calculate the approximate deviation of the trajectory arc: multiply the slope by the divergence average (divergence average = (0.3+0.4+0.5) / 3=0.4, unitless), and get the deviation 0.1×0.4=0.04; then repeat the calculation for each divergence value, the deviation vector is (0.1×0.3=0.03, 0.1×0.4=0.04, 0.1×0.5=0.05), and the total deviation average is 0.04; the calculation logic is to first determine the slope and then multiply and average point by point to ensure that the deviation reflects the divergence intensity. In the experiment, the divergence data is (0.3, 0.4, 0.5) (unitless), the singularity slope is determined to be 0.1 per segment, and the deviation is calculated to be 0.04 (unitless), the underlying parameter segment interval is 1, and the number of points is 3. The calculation is completed through difference and multiplication operations, for example, the slope formula , where s is the slope of the singular point 0.1 per segment, is the last divergence value 0.5, is the first divergence value 0.3, n is the number of segments 3; the deviation formula di=s⋅fi, where di is the i-th deviation value (such as 0.03, unitless), fi is the i-th divergence value (such as 0.3, unitless), and the total deviation is the average of all di.

[0030] The average values ​​of the state parameters extracted from step S231 (yaw rate of change 0.8 radians per second squared (rad / s²), yaw rate of change approximately 0.174 radians per second (rad / s²), and acceleration rate of change 1 meter per second cubed (m / s³)) are derived, and the deviation 0.04 (unitless) is extracted from step S233. The derivation is achieved by multiplying the parameter weights by the deviations, with the weights set to 0.5 (unitless) for yaw, 0.3 (unitless) for yaw, and 0.2 (unitless) for acceleration, with the total weight being 1 (unitless). First, calculate the weighted sum: yaw parameter weight = 0.8 rad / s² × 0.5 = 0.4, in rad / s²; yaw parameter weight = 0.174 rad / s² × 0.3 = 0.0522, in rad / s²; acceleration parameter weight = 1 m / s³ × 0.2 = 0.2, in m / s³. The total weighted sum is 0.4 + 0.0522 + 0.2 = 0.6522 (combined units: Due to different parameter units, this is the weighted combination of physical quantities. After multiplying by the unitless bias, the degree of disorder is unitless). Then, the degree of disorder = total weighted sum × bias = 0.6522 × 0.04 = 0.0261. The process is serialized derivation. If the parameters are sequential (for example, the yaw sequence (0.7, 0.8, 0.9) in rad / s², the roll sequence (0.15, 0.174, 0.198) in rad / s, and the acceleration sequence (0.9, 1, 1.1) in m / s³), and the corresponding deviation sequence (0.03, 0.04, 0.05), then the calculation logic for each point is: first calculate the weighted sum of the parameters at that point, and then multiply it by the corresponding deviation. For example, point 1: (0.7×0.5+0.15×0.3+0.9×0.2)×0.03=(0.35+0.045+0.18)×0.03=0.575×0.03≈0.0173 (no unit); point 2: (0.8×0.5+0.174×0.3+1×0.2)×0.04≈(0.4+0.0522+0.2)×0.04≈0.6522× 0.04≈0.0261 (unitless); Point 3: (0.9×0.5+0.198×0.3+1.1×0.2)×0.05=(0.45+0.0594+0.22)×0.05=0.7294×0.05≈0.0365 (unitless); the final degree of disorder is the average of the three points, that is, (0.0173+0.0261+0.0365) / 3≈0.0266.

[0031] Step S24 includes the following steps: Step S241: Decomposing the disordered motion posture of the phase trajectory to obtain a trajectory decomposition vector; Step S242: performing kernel density estimation of potential collision points on the obstacle orientation according to the trajectory decomposition vector to generate probability distribution data of potential collision points; Step S243: performing time series approximate variance constraints on the dynamic parameters of the trajectory decomposition vector to obtain dynamic parameter constraint variances; wherein the dynamic parameters include the vehicle body trajectory motion yaw rate, sideslip angle, and acceleration; Step S244: performing lateral swing curvature correlation deduction based on the potential collision point probability distribution data and the dynamic parameter constraint variance to obtain the trajectory swing curvature associated with the collision point; Step S245: performing collision posture trajectory analysis based on the power parameter constraint variance and the trajectory swing curvature associated with the collision point to generate a collision posture trajectory.

[0032] In an embodiment of the present invention, the principal component analysis method is used to decompose the disordered motion posture of the phase trajectory disorder degree. First, a three-dimensional orthogonal decomposition space is constructed, and the axes correspond to yaw motion, sideslip motion, and vertical motion respectively. A posture matrix is ​​established, and the matrix elements are composed of 200 sampling points of the phase plane trajectory, and each point contains position and velocity information. The posture matrix is ​​subjected to singular value decomposition to calculate the eigenvalues ​​and eigenvectors. The three main eigenvalues ​​are measured to be 25.8, 15.4, and 8.6, respectively, and the corresponding eigenvectors are [0.82, 0.42, 0.38], [0.35, 0.75, 0.55], and [0.45, 0.51, 0.73]. The contribution rate of each principal component is calculated based on the size of the eigenvalue, which is 51.6%, 30.8%, and 17.6%, respectively. The phase trajectory disorder degree of 0.5 is distributed to the three eigenvectors according to the contribution rate, and the trajectory decomposition vector [0.258, 0.154, 0.088] is obtained. This vector represents the degree of disorder in the three dimensions of yaw motion, sideslip motion and vertical motion, with the yaw motion component being the largest and the vertical motion component being the smallest.

[0033] The kernel density of potential collision points is estimated for the obstacle orientation based on the trajectory decomposition vector [0.258, 0.154, 0.088]. The obstacle is a stationary pedestrian located 28 meters in front of the vehicle and offset 1.5 meters laterally. First, a set of possible vehicle trajectories is generated. 500 random trajectories are generated using the Monte Carlo method. Each trajectory contains 20 points with a time interval of 0.1 seconds, covering the next 2 seconds. Trajectory generation takes into account the disorder component in the trajectory decomposition vector. The yaw component affects the steering angle change, the sideslip component affects the tire lateral force, and the vertical component affects the suspension response. A 30×10 meter rectangular area is constructed in front of the vehicle with a grid size of 0.5×0.5 meters, totaling 1200 grid points. A two-dimensional Gaussian kernel function is applied to each grid point to calculate the collision probability density. The kernel bandwidth matrix is ​​[1.8m, 0.0m; 0.0m, 0.9m]. The collision probability density calculation formula is: , where K is the Gaussian kernel function, ( , ) are the coordinates of the trajectory points, and is the bandwidth parameter (1.8m and 0.9m), and n is the total number of trajectory points. The calculation results show that the area with the highest collision probability is within 1.5m around the obstacle, with the maximum probability density value of 0.28.

[0034] The trajectory decomposition vector [0.258, 0.154, 0.088] is subjected to the temporal approximate variance constraints between the dynamic parameters. First, the physical limit constraints of the dynamic parameters are set: the yaw rate limit is ±30° / s, which is converted to the international standard unit as ±30×π / 180≈±0.5236 (rad / s); the sideslip angle limit is ±12 degrees, which is converted to the international standard unit as ±12×π / 180≈±0.2094 (rad); the lateral acceleration limit is ±10m / s². Combining the vehicle state equation and the trajectory decomposition vector, the noise covariance matrix Q is constructed. The diagonal elements of the noise covariance matrix are the squares of the trajectory decomposition vector components multiplied by the scale factor 15. The specific calculation is: the first element is 0.258²×15≈0.066564×15≈0.99846, the second element is 0.154²×15≈ 0.023716×15≈0.35574, and the third element 0.088²×15≈0.007744×15≈0.11616. The state covariance matrix is ​​then solved using the discrete Lyapunov equation. After 10 iterations, the steady-state covariance matrix P is obtained. The variance of the dynamic parameters is extracted from P: the yaw rate variance is 28.5°², which is converted to 28 in international standard units. 5×(π / 180)²≈28.5×(0.0174533)²≈28.5×0.0003046≈0.008681 (rad² / s²); the sideslip angle variance is 6.8°², which is converted to international standard units as 6.8×(π / 180)²≈6.8×0.0003046≈0.002071 (rad²); the lateral acceleration variance is 3.2m² / s 4 ,These variance values ​​characterize the fluctuation range of each dynamic parameter under the influence of ,the disordered trajectory decomposition vector, and provide constraints for subsequent path ,planning.

[0035] The curvature-probability correlation analysis method is used to deduce the lateral swing curvature correlation based on the probability distribution data of potential collision points and the variance of power parameter constraints. First, the curvature of the path from the current position of the vehicle to each potential collision point is calculated. The curvature calculation formula is: Where (x0, y0) is the current position of the vehicle, ( , ) are the coordinates of the potential collision point. For the area with the highest collision probability (probability density 0.28), the coordinates are approximately (28, 1.5), in meters. The calculated path curvature is 0.0375 / m. Considering the impact of the dynamic parameter constraint variance on curvature, a yaw rate variance of 28.5 rad² / s² corresponds to a curvature variance of 0.000625 / m², a sideslip angle variance of 6.8 rad² corresponds to a curvature variance of 0.000484 / m², and a lateral acceleration variance of 3.2 m² / s². 4 The corresponding curvature variance is 0.000324 / m². These curvature variances are weighted and summed with weights of 0.5, 0.3 and 0.2 respectively, resulting in a comprehensive curvature variance of 0.000518 / m². The reference curvature of 0.0375 / m is added to the curvature standard deviation of 0.0228 / m to obtain the trajectory swing curvature of 0.0603 / m associated with the collision point. Based on the dynamic parameter constraint variance and the trajectory swing curvature associated with the collision point, a collision posture trajectory analysis is performed to generate a collision posture trajectory. In the embodiment, a collision posture trajectory analysis is performed based on the dynamic parameter constraint variance and the trajectory swing curvature of 0.0603 / m associated with the collision point. A time step of 0.1s and a total calculation time of 2.5s are used, and a total of 25 trajectory points are generated. The initial state is set as: position (0m, 0m), speed 15.5m / s, heading angle 0rad, yaw angular velocity 2.5rad / s, and sideslip angle 1.8rad. At each time step, the next state is predicted using a state equation that includes updates to the yaw rate, sideslip angle, and position. To account for the variance of the power parameter constraints, a random perturbation is added to each prediction step. The perturbation follows a normal distribution with a mean of 0 and a variance equal to the variance of the power parameter constraints. The trajectory curvature of 0.0603 / m translates to a path radius of 16.6m, limiting the vehicle's steering ability. Simulation results show that the vehicle reaches a lateral displacement of 1.4m, a yaw rate of 18.5rad / s, and a sideslip angle of 6.5rad at 1.7s. At 2.1s, the vehicle collides with the obstacle at a collision position of (28m, 1.5m), with a velocity of 13.8m / s and a heading angle of 23rad. The resulting collision trajectory fully records the vehicle's motion from the current state to the collision state.

[0036] Step S3 includes the following steps: Step S31: performing feature learning on the collision posture prediction trajectory to obtain a collision prediction feature trajectory; Step S32: performing self-attention weight aggregation on the collision prediction feature trajectory to obtain a collision prediction weight trajectory; Step S33: performing emergency avoidance safety braking parameter matching on the vehicle braking process state according to the collision prediction weight trajectory and the obstacle orientation to obtain avoidance braking safety matching parameters; Step S34: performing collision avoidance path planning learning based on the avoidance braking safety matching parameters and the collision prediction weight trajectory to generate a collision avoidance planning path.

[0037] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes: Step S31: performing feature learning on the collision posture prediction trajectory to obtain a collision prediction feature trajectory; In the embodiment of the present invention, the collision posture prediction trajectory generated in step S2, which consists of vehicle posture data points {abscissa, ordinate, yaw angle, body slip angle} at 100 consecutive time points (time interval of 0.02s) in the next 2s, is input into a preset sliding window analysis program. The program sets the window width to 10 data points and the sliding step to 5 data points. In each window, the three physical quantities of the maximum curvature, the peak yaw rate and the mean slip angle change rate of the trajectory segment are extracted. For example, for the 10 data points in the first window, the calculated curvature is 0.05 , the peak yaw rate is 15 rad / s, and the mean slip angle rate is 3 rad / s. After completing the calculation for all windows, the resulting 19 sets of {maximum curvature, peak yaw rate, mean slip angle rate} feature vector sequences are output as collision prediction feature trajectories.

[0038] Step S32: performing self-attention weight aggregation on the collision prediction feature trajectory to obtain a collision prediction weight trajectory; In an embodiment of the present invention, the 19 groups of feature vector sequences generated in step S31 are processed. The 10th group of feature vectors is taken as the target vector for the current analysis. The dot product similarity score between the target vector and all 19 feature vectors including itself is calculated. For example, the dot product scores of the 10th group of vectors with the 9th, 10th, and 11th groups of vectors (representing the period of rapid instability) are 0.92, 0.99, and 0.91, respectively, while the scores with the 1st and 2nd groups of vectors (representing the early stage of instability) are 0.35 and 0.38, respectively. These 19 scores are normalized to obtain a set of weight coefficients, the sum of which is 1. This set of weight coefficients is used to perform weighted summation on the original 19 groups of feature vectors to generate a new "weighted aggregated" feature vector. Repeat this process for all 19 groups of feature vectors, and finally obtain a new sequence consisting of 19 "weighted aggregated" feature vectors, i.e., the collision prediction weight trajectory.

[0039] Step S33: performing emergency avoidance safety braking parameter matching on the vehicle braking process state according to the collision prediction weight trajectory and the obstacle orientation to obtain avoidance braking safety matching parameters; In this embodiment of the present invention, braking parameter matching is performed based on the highest-risk feature vector (e.g., a yaw rate peak reaching 30 rad / s) in the collision prediction weighted trajectory generated in step S32, combined with the obstacle's position (12 meters directly ahead). First, a safety distance threshold of 2 meters is set. Based on the vehicle's current speed of 20 m / s, the minimum average deceleration required to maintain a 2-meter safety distance is calculated to be 8.3 m / s². Using this deceleration as a target, a search is performed within a pre-set brake pressure distribution map. This map records the vehicle response under different combinations of {inner rear wheel brake pressure, outer front wheel brake pressure}. The resulting set of parameters is: 6.1 MPa for the inner rear wheel brake pressure and 8.2 MPa for the outer front wheel brake pressure. This combination achieves a deceleration of at least 8.3 m / s² without increasing yaw instability. This set of pressure values ​​{6.1 MPa, 8.2 MPa} serves as the avoidance braking safety matching parameters.

[0040] Step S34: performing collision avoidance path planning learning based on the avoidance braking safety matching parameters and the collision prediction weight trajectory to generate a collision avoidance planning path.

[0041] In this embodiment of the present invention, path generation is performed based on the avoidance braking safety matching parameters {inner rear wheel brake pressure: 6.1 MPa, outer front wheel brake pressure: 8.2 MPa} obtained in step S33 and the collision prediction weighted trajectory. The vehicle's current state (position, speed 20 m / s, attitude angle, etc.) and the braking parameters are input into a vehicle dynamics forward simulation program. This program iteratively calculates the vehicle's motion over the next 2.5 seconds with a step size of 0.01 seconds. At each step size, the program calculates the wheel slip and tire force based on the braking pressure, then solves for the vehicle body force and torque, updates the vehicle's linear acceleration and angular acceleration, and finally integrates to obtain the velocity and position at the next moment. The sequence of vehicle center of mass coordinate points (x, y) at these 250 consecutive moments is connected to form a precise, physically feasible trajectory, which is the generated collision avoidance planning path.

[0042] Step S33 includes the following steps: Step S331: Matching safety braking parameters between changes in obstacle distances to the vehicle braking process state based on the collision prediction weight trajectory and the obstacle orientation to obtain distance-related safety braking parameters; wherein the distance-related safety braking parameters include: distance-related safety braking speed, time, and wheel locking threshold; Step S332: Based on the distance-related safety braking speed, time, and wheel lock threshold in the distance-related safety braking parameters, the collision prediction weight trajectory and the vehicle braking process state are subjected to time-series distribution of steering braking force to the inner rear wheel and the outer front wheel to obtain steering braking force distribution data; Step S333: performing torque vector regression analysis coupling based on the steering braking force distribution data to obtain a torque coupling vector; Step S334: performing vehicle trajectory stabilization posture control on the collision prediction weight trajectory and the vehicle braking process state according to the steering braking force distribution data and the torque coupling vector to obtain vehicle trajectory stabilization posture control data; Step S335: performing emergency avoidance safety braking parameter matching based on the vehicle body stability posture control data, the steering braking force distribution data, and the torque coupling vector to obtain avoidance braking safety matching parameters.

[0043] In this embodiment of the present invention, the vehicle's current braking state (vehicle speed of 25 m / s) and the obstacle's position (40 m directly ahead) are obtained. Based on the collision prediction weighted trajectory, the actual path length to the obstacle is calculated to be 42 m. A fixed, inviolable final safety distance of 2 m is set, resulting in the vehicle being required to come to a complete stop within a path distance of 40 m (42 m - 2 m). Based on the kinematic relationship a = v² / 2d, the target average deceleration required to come to a complete stop within this distance is calculated as a = (25 m / s²)² / (2 × 40 m) = 7.8125 m / s², expressed in m / s². Based on this deceleration and the initial speed, the time required to complete braking is calculated as t = v / a = 25 m / s² / 7.8125 m / s² = 3.2 seconds. Based on the vehicle's dynamic parameters and the current road adhesion coefficient (e.g., 0.8), the maximum wheel slip threshold required to achieve this deceleration without causing complete wheel lock is determined to be 18%. Finally, the distance-related safety braking parameters obtained are {distance-related safety braking speed: 25 m / s, time: 3.2 s, wheel locking threshold: 18% slip rate}.

[0044] Based on the obtained distance-dependent safety braking parameters (braking speed 25 m / s, time 3.2 s, locking threshold 18%), the braking force for the next 3.2 s is temporally distributed. The 3.2 s braking process is discretized into 32 time steps of 0.1 s each. At each time step, the yaw moment and total longitudinal braking force required to achieve trajectory tracking are calculated based on the target curvature determined by the collision prediction weighted trajectory and the vehicle model (vehicle mass 1600 kg, wheelbase 2.8 m). For example, at 1.5 s, the vehicle needs to generate a total braking force of -12000 N and a yaw moment of 500 N·m. By solving the mechanical equations, this requirement is distributed to the outer front wheel and inner rear wheel on the steering side. The calculated braking force is 7357 N for the outer front wheel and 4643 N for the inner rear wheel. The wheel slip ratio corresponding to the calculated braking force is monitored to ensure it does not exceed the 18% threshold. The sequence of {time point, outer front wheel braking force, inner rear wheel braking force} calculated in all 32 time steps constitutes the steering braking force distribution data.

[0045] The steering braking force distribution data generated in step S332 is converted to actuator torque. For example, the data at 1.5 seconds shows the required braking force for the outer front wheel is 7357 N, the required braking force for the inner rear wheel is 4643 N, and the dynamic wheel radius is 0.32 m. The braking torque for the outer front wheel is achieved by the hydraulic braking system, with the required torque being 7357 N x 0.32 m = 2354.24 N·m. The inner rear wheel is braked by both the drive motor and the hydraulic brake. A pre-set two-dimensional lookup table, indexed by vehicle speed and required braking force, is queried to determine the optimal distribution of the motor's regenerative braking torque and the hydraulic braking torque. At the current vehicle speed (for example, 15 m / s), the motor can provide a maximum regenerative braking torque of 1200 N·m (corresponding to a braking force of 3750 N). The remaining 893 N of braking force (4643 N - 3750 N) is provided by the hydraulic brake, resulting in a torque of 893 N × 0.32 m = 285.76 N·m. The torque distribution at this moment is recorded as {outer front wheel hydraulic pressure: 2354.24, inner rear wheel regenerative: 1200, inner rear wheel hydraulic pressure: 285.76} (in N·m). This is the torque coupling vector at that moment.

[0046] It runs as a closed-loop feedback controller with an execution cycle of 20ms. At 1.52s, the actual yaw rate of the vehicle obtained by the on-board inertial measurement unit (IMU) and GPS is 12.5rad / s, and the side slip angle of the body is -2.1rad. The reference value of the collision prediction weight trajectory at this moment is: yaw rate 12.0rad / s, side slip angle -2.0rad. The calculated errors are: yaw rate error e_ω=12.5-12.0=0.5, in rad / s, and side slip angle error e_β=-2.1-(-2.0)=-0.1, in rad. The error is input into a preset proportional-integral-derivative (PID) controller with parameters Kp=80, Ki=10, and Kd=5. The controller outputs a corrected yaw moment , calculated = 42 N·m. This correction torque will be used to fine-tune the braking force distribution at the next moment. This series of data streams calculated over time ({time point, yaw rate error, sideslip angle error, corrected yaw moment}) constitutes the vehicle body stability control data.

[0047] In each control cycle (20ms), the previous calculation results are integrated to generate the final actuator control command. Taking the control cycle of 1.52s as an example, the basic braking force distribution command at that moment is first interpolated from the steering braking force distribution data in step S332. Then, the corrected yaw moment is obtained from the vehicle body stability control data in step S334. =42N·m. To achieve this correction torque, an equal and opposite force ΔF= is applied to the inner and outer wheels. / wheelbase, ΔF = 42 N·m / 1.6m = 26.25, expressed in N. This correction force is added to the base braking force: the outer front wheel's braking force is reduced by 26.25 N, while the inner rear wheel's braking force is increased by 26.25 N. Finally, the corrected braking force is applied through the torque coupling conversion logic described in step S333 to calculate the final motor regenerative braking torque and hydraulic brake disc brake pressure commands. This final output, {motor target torque, master cylinder target pressure}, integrating feedforward and feedback control, serves as the evasive braking safety matching parameters at that moment.

[0048] Step S34 includes the following steps: Step S341: performing a weight analysis of the traversable space of the avoidance path based on the avoidance braking safety matching parameters and the collision prediction weight trajectory to obtain the traversable space weight of the avoidance path; Step S342: performing an iterative simulation of the controllability of vehicle braking parameters associated with the traversable space weight of the avoidance path according to the avoidance braking safety matching parameter to obtain a controllable braking parameter data set on the spatial path; Step S343: performing optimal solution decision on the controllable parameters of the controllable braking parameter data set to obtain optimal controllable solution parameters; Step S344: performing collision avoidance path planning learning based on the avoidance path traversable space weight and controllable optimal solution parameters to generate a collision avoidance planning path.

[0049] In one scenario of this embodiment, based on the avoidance braking safety matching parameters {inner rear wheel brake pressure: 6.1 MPa, outer front wheel brake pressure: 8.2 MPa} and the collision prediction weight trajectory, a grid map of the forward range of 40 m × 12 m is constructed, with a grid size of 0.2 m × 0.2 m. A traffic weight is calculated for each grid cell, with an initial value of 1.0. The obstacle area (the coordinate point set {(25,2), (26,2), (25,3), (26,3)}), in meters, has a weight of 0. The weight within 1.5 m of the obstacle decreases with distance: specifically, the weight within 0.5 m is 0.1, 0.5-1 m is 0.3, and 1-1.5 m is 0.6. The weight of the area with a ground friction coefficient of 0.8 remains unchanged, while the weight of the area with a friction coefficient of 0.6 (coordinate range x∈[15,20], y∈[0,4]) is reduced by 0.25. The collision prediction weight is increased within 1.8 meters on either side of the trajectory. Within 0.5 meters, the weight increases by 0.4, within 0.5-1 meters, by 0.3, and between 1 and 1.8 meters, by 0.15. The weight is reduced by 0.5 in the boundary region (y < 0.8 or y > 11.2). A 5×5 Gaussian convolution kernel is used for smoothing to obtain the spatial weight matrix for the avoidance path.

[0050] Based on the avoidance braking safety matching parameters, 300 candidate paths were generated within a grid space with a weight greater than 0.7, consisting of 30 path points and a spacing of 1.2 meters. Vehicle parameters were set as follows: mass 1680 kg, wheelbase 2.78 m, front and rear track 1.58 m, center of mass height 0.54 m, and tire radius 0.31 m. For each path, an iterative simulation of braking parameter controllability was performed with a step size of 0.01 seconds for a total of 2.5 seconds. Initial conditions were: vehicle speed 20 m / s, heading angle 0 degrees, slip angle 0 degrees, and yaw rate 0 rad / s. Each iteration applied a brake pressure of 6.1 MPa to the inner rear wheel and 8.2 MPa to the outer front wheel. Tire forces were calculated: longitudinal force was obtained from a slip ratio-force curve table, with a maximum static friction of 5600 N; lateral force was obtained from a slip angle-force curve table, with a maximum lateral force of 4800 N. During the iteration, paths with wheel slip exceeding 0.18 or slip angle exceeding 7.5 degrees were deemed uncontrollable. For each controllable path, the position, velocity, acceleration, and brake pressure changes were recorded to form a controllable braking parameter dataset.

[0051] The optimal solution was determined for 175 paths from a dataset with controllable braking parameters. An evaluation function was constructed, consisting of four metrics: safety distance D (minimum distance to obstacles, target value greater than 3 meters), trajectory smoothness S (maximum curvature, target value less than 0.22 / m), braking efficiency E (braking distance, target value less than 30 meters), and ride comfort C (maximum resultant acceleration, target value less than 4.5 m / s²). The evaluation function is F = 0.35 × D + 0.25 × S + 0.25 × E + 0.15 × C, with a maximum score of 100. Each path was substituted into the vehicle dynamics equation to calculate the values ​​of each metric, which was solved using the fourth-order Runge-Kutta method. The metrics were normalized to a scale of 0-100. For example, a safety distance of 3 meters would receive 60 points, while a curvature of 0.22 / m would receive 60 points, while a curvature of 0.15 / m would receive 100 points. The comprehensive scores of the 175 paths were calculated, and the parameters corresponding to the path with the highest score were selected, which were the controllable optimal solution parameters, including a sequence of brake pressure and steering angle at 30 discrete points.

[0052] A collision avoidance plan is generated based on the traversable space weights and controllable optimal solution parameters. First, the controllable optimal solution parameters are used as initial seeds and perturbed by ±3% in both the brake pressure and steering angle dimensions, generating 20 sets of variant parameters. Each set of parameters is simulated forward using the vehicle dynamics equations with a step size of 0.01s and a duration of 2.5s. The dynamics equations include body dynamics, tire dynamics, and steering system dynamics, totaling 12 state variables. Double-precision floating-point numbers are used for accuracy. A path integration is performed on the 20 trajectories and the traversable space weight matrix of the avoidance path to calculate the total weighted score. The trajectory with the highest weighted score is selected and smoothed using cubic spline interpolation with a sampling interval of 0.1m. This path meets the requirements of a maximum lateral acceleration of no more than 3.8m / s², a maximum longitudinal deceleration of no more than 6.5m / s², and a steering angle change rate of no more than 6 degrees / s. The final output is a collision avoidance plan consisting of 250 time points, each containing position, velocity, acceleration, and attitude angle information.

[0053] The present invention also provides an intelligent vehicle path planning system for executing the intelligent vehicle path planning method described above. The intelligent vehicle path planning system includes: A data extraction module is used to extract the obstacle position and vehicle braking status during the intelligent vehicle's emergency avoidance process from the intelligent vehicle control center; The collision trajectory prediction module is used to perform attitude lateral phase plane instability analysis on the vehicle braking process state to obtain the attitude phase plane instability state; based on the attitude phase plane instability state and the obstacle orientation, the collision attitude trajectory is analyzed to generate the collision attitude trajectory; The avoidance path planning learning module is used to match the emergency avoidance safety braking parameters of the vehicle braking process state according to the collision posture prediction trajectory to obtain the avoidance braking safety matching parameters; based on the avoidance braking safety matching parameters and the collision posture prediction trajectory, collision avoidance path planning learning is performed to generate a collision avoidance planning path.

[0054] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A method for intelligent vehicle path planning, characterized in that: The following steps are involved: Step S1: extracting the obstacle orientation and vehicle braking process status during the emergency avoidance process of the intelligent vehicle from the intelligent vehicle control center; Step S2: performing a posture lateral phase plane instability analysis on the vehicle braking process state to obtain a posture phase plane instability state; performing a collision posture trajectory analysis based on the posture phase plane instability state and the obstacle orientation to generate a collision posture trajectory; Step S3: matching the emergency avoidance safety braking parameters to the vehicle braking process state according to the collision posture prediction trajectory to obtain the avoidance braking safety matching parameters; Based on the avoidance braking safety matching parameters and the collision posture prediction trajectory, collision avoidance path planning learning is performed to generate a collision avoidance planning path.

2. The intelligent vehicle path planning method according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: performing out-of-control speed fluctuation analysis on the vehicle braking process state to obtain out-of-control fluctuation speed; Step S22: performing attitude lateral phase plane instability analysis on the vehicle braking process state according to the out-of-control fluctuation speed, thereby obtaining an attitude phase plane instability state; Step S23: deriving the disorder degree of the phase trajectory of the attitude phase plane instability state; Step S24: performing collision posture trajectory analysis based on the phase trajectory disorder degree and obstacle orientation to generate a collision posture trajectory.

3. The intelligent vehicle path planning method according to claim 2, characterized in that: Step S22 includes the following steps: Step S221: Obtain basic information of the smart car; wherein the basic information includes the body structure, body weight distribution and suspension design structure of the smart car; Step S222: extracting the braking-steering ratio in the vehicle braking process; performing numerical integration of the steering centrifugal torque on the intelligent vehicle body structure and body weight distribution in the basic information according to the braking-steering ratio and the uncontrolled fluctuation speed to obtain steering centrifugal torque increment data; Step S223: analyzing the incremental changes in caster angle and camber angle of the suspension design structure through the steering centrifugal torque incremental data; Step S224: deriving the vehicle body pitch angle increment gradient based on the steering centrifugal torque increment data and the change increments of the caster angle and the camber angle to obtain the attitude increment pitch angle; Step S225: performing attitude lateral phase plane instability analysis based on the steering centrifugal torque increment data, the caster angle, the camber angle change increment and the attitude increment pitch angle, thereby obtaining an attitude phase plane instability state.

4. The intelligent vehicle path planning method according to claim 2, characterized in that: Step S23 includes the following steps: Step S231: extracting the vehicle instability offset trajectory in the attitude phase plane instability state to construct an offset phase trajectory plane diagram; extracting state parameters in the attitude phase plane instability state, wherein the state parameters include the yaw rate change rate, the vehicle sideslip angle change rate, and the lateral acceleration change rate; Step S232: performing trajectory spiral divergence analysis on the offset phase trajectory plane diagram to obtain trajectory spiral divergence data; Step S233: determining the trajectory singularity slope based on the trajectory spiral divergence data; calculating the trajectory arc approximate deviation based on the trajectory singularity slope and the trajectory spiral divergence data; Step S234: Deducing the phase trajectory disorder degree of the trajectory arc approximation deviation according to the yaw rate change rate, the vehicle body sideslip angle change rate and the lateral acceleration change rate in the state parameters to obtain the phase trajectory disorder degree.

5. The intelligent vehicle path planning method according to claim 2, characterized in that: Step S24 includes the following steps: Step S241: Decomposing the disordered motion posture of the phase trajectory to obtain a trajectory decomposition vector; Step S242: performing kernel density estimation of potential collision points on the obstacle orientation according to the trajectory decomposition vector to generate probability distribution data of potential collision points; Step S243: performing time series approximate variance constraints on the dynamic parameters of the trajectory decomposition vector to obtain dynamic parameter constraint variances; wherein the dynamic parameters include the vehicle body trajectory motion yaw rate, sideslip angle, and acceleration; Step S244: performing lateral swing curvature correlation deduction based on the potential collision point probability distribution data and the dynamic parameter constraint variance to obtain the trajectory swing curvature associated with the collision point; Step S245: performing collision posture trajectory analysis based on the power parameter constraint variance and the trajectory swing curvature associated with the collision point to generate a collision posture trajectory.

6. The intelligent vehicle path planning method according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: performing feature learning on the collision posture prediction trajectory to obtain a collision prediction feature trajectory; Step S32: performing self-attention weight aggregation on the collision prediction feature trajectory to obtain a collision prediction weight trajectory; Step S33: performing emergency avoidance safety braking parameter matching on the vehicle braking process state according to the collision prediction weight trajectory and the obstacle orientation to obtain avoidance braking safety matching parameters; Step S34: performing collision avoidance path planning learning based on the avoidance braking safety matching parameters and the collision prediction weight trajectory to generate a collision avoidance planning path.

7. The intelligent vehicle path planning method according to claim 6, characterized in that: Step S33 includes the following steps: Step S331: Matching safety braking parameters between changes in obstacle distances to the vehicle braking process state based on the collision prediction weight trajectory and the obstacle orientation to obtain distance-related safety braking parameters; wherein the distance-related safety braking parameters include: distance-related safety braking speed, time, and wheel locking threshold; Step S332: Based on the distance-related safety braking speed, time, and wheel lock threshold in the distance-related safety braking parameters, the collision prediction weight trajectory and the vehicle braking process state are subjected to time-series distribution of steering braking force to the inner rear wheel and the outer front wheel to obtain steering braking force distribution data; Step S333: performing torque vector regression analysis coupling based on the steering braking force distribution data to obtain a torque coupling vector; Step S334: performing vehicle trajectory stabilization posture control on the collision prediction weight trajectory and the vehicle braking process state according to the steering braking force distribution data and the torque coupling vector to obtain vehicle trajectory stabilization posture control data; Step S335: performing emergency avoidance safety braking parameter matching based on the vehicle body stability posture control data, the steering braking force distribution data, and the torque coupling vector to obtain avoidance braking safety matching parameters.

8. The intelligent vehicle path planning method according to claim 6, characterized in that: Step S34 includes the following steps: Step S341: performing a weight analysis of the traversable space of the avoidance path based on the avoidance braking safety matching parameters and the collision prediction weight trajectory to obtain the traversable space weight of the avoidance path; Step S342: performing an iterative simulation of the controllability of vehicle braking parameters associated with the traversable space weight of the avoidance path according to the avoidance braking safety matching parameter to obtain a controllable braking parameter data set on the spatial path; Step S343: performing optimal solution decision on the controllable parameters of the controllable braking parameter data set to obtain optimal controllable solution parameters; Step S344: performing collision avoidance path planning learning based on the avoidance path traversable space weight and controllable optimal solution parameters to generate a collision avoidance planning path.

9. An intelligent vehicle path planning system, characterized in that: For executing the intelligent vehicle path planning method according to claim 1, the intelligent vehicle path planning system comprises: A data extraction module is used to extract the obstacle position and vehicle braking status during the intelligent vehicle's emergency avoidance process from the intelligent vehicle control center; The collision trajectory prediction module is used to perform attitude lateral phase plane instability analysis on the vehicle braking process state to obtain the attitude phase plane instability state; based on the attitude phase plane instability state and the obstacle orientation, the collision attitude trajectory is analyzed to generate the collision attitude trajectory; The avoidance path planning learning module is used to match the emergency avoidance safety braking parameters of the vehicle braking process state according to the collision posture prediction trajectory to obtain the avoidance braking safety matching parameters; based on the avoidance braking safety matching parameters and the collision posture prediction trajectory, collision avoidance path planning learning is performed to generate a collision avoidance planning path.

Images