Transverse and longitudinal composite collision avoidance control method for electric automobile

Abstract

The invention provides a transverse and longitudinal composite collision avoidance control method for an electric vehicle, which comprises the following steps of: performing transverse and longitudinal track decoupling planning under a Frenet coordinate system based on information of a vehicle, a barrier vehicle and a road to obtain transverse and longitudinal candidate track clusters; performing axial independent constraint optimization on the transverse and longitudinal candidate track clusters by adopting an improved Lattice algorithm, performing combined track feasibility constraint optimization on the optimized transverse and longitudinal track clusters, and screening out an optimal collision avoidance track from the track clusters conforming to constraints through a target function; on the basis of the optimal collision avoidance trajectory, transverse and longitudinal composite collision avoidance control is achieved through a decoupling trajectory tracking controller, speed-position error closed-loop adjustment is achieved through a double-loop PID controller in longitudinal control, and an MPC tracker is adopted in transverse control. According to the invention, the feasibility of the trajectory is preferentially guaranteed by using a hierarchical constraint mechanism in the trajectory planning module, and the collision avoidance capability of the automobile in a high-speed low-adhesion scene is enhanced by using a decoupling control architecture.

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicles, and in particular to a transverse and longitudinal composite collision avoidance control method for an electric vehicle. Background Art

[0002] With the popularization and intelligent development of electric vehicles, vehicle active safety technology has become a focus of industry research. Among them, vehicle emergency braking systems and emergency steering collision avoidance systems, as typical active safety technologies, play a vital role in preventing road traffic accidents. However, under low-adhesion road conditions such as ice, snow, and wet conditions, traditional collision avoidance systems based on single braking or steering control logic face severe technical challenges. Conventional AEB systems rely on braking and deceleration to avoid collisions, but on low-adhesion roads, the friction coefficient between the tires and the ground is significantly reduced, and the braking distance is greatly increased. Relying solely on longitudinal braking may lead to collision avoidance failure; and existing lateral lane change collision avoidance methods adjust the path through steering, but under high-speed, low-adhesion conditions, the vehicle's lateral dynamic stability deteriorates sharply, and steering alone can easily cause skidding or even loss of control.

[0003] Trajectory planning and tracking control: Existing motion planning algorithms are mostly designed for conventional scenarios. This leads to redundant candidate trajectory generation and low computational efficiency in emergency collision avoidance. Furthermore, tracking control often employs coupled control, which, while promising global optimization potential, requires the introduction of additional system parameters and control variables, significantly increasing the complexity of control system design and debugging.

[0004] Therefore, there is an urgent need for a composite lateral and longitudinal collision avoidance control method for electric vehicles to achieve efficient coordination between vehicle braking and steering, and to break through the limitations of a single collision avoidance control strategy. Summary of the Invention

[0005] In view of this, the present invention provides a method for controlling a lateral and longitudinal composite collision avoidance of an electric vehicle, comprising: Based on the information of the ego vehicle, obstacle vehicles and road, the horizontal and vertical trajectory decoupling planning is performed in the Frenet coordinate system to obtain the horizontal and vertical candidate trajectory clusters; An improved Lattice algorithm is used to perform axial independent constraint optimization on the lateral and longitudinal candidate trajectory clusters. The optimized lateral and longitudinal trajectory clusters are then subjected to combined trajectory feasibility constraint optimization. The trajectory clusters that meet the constraints are screened out as the optimal collision avoidance trajectory through the objective function. Based on the optimal collision avoidance trajectory, a decoupled trajectory tracking controller is used to implement lateral and longitudinal composite collision avoidance control, wherein a dual-loop PID controller is used for longitudinal control to achieve closed-loop adjustment of speed-position error, and an MPC tracker is used for lateral control.

[0006] As a further illustration of the present invention, the lateral and longitudinal candidate trajectory clusters are obtained by performing lateral and longitudinal trajectory decoupling planning in the Frenet coordinate system based on the information of the ego vehicle, the obstacle vehicle, and the road, specifically including: The pose parameters of the ego vehicle and the obstacle vehicle are converted from the Cartesian coordinate system to the Frenet coordinate system with the road centerline as the reference; The lateral and longitudinal trajectories of the ego vehicle are fitted with quintic polynomials to generate the LS and St curves, respectively. The St curve represents the longitudinal displacement of the ego vehicle described by a quintic polynomial with time t as the independent variable; the LS curve represents the lateral displacement described by a quintic polynomial with the longitudinal displacement S as the independent variable. By setting the starting point and end point position information of the horizontal track and the vertical track, and generating the horizontal and vertical track clusters according to the LS curve and the St curve.

[0007] As a further illustration of the present invention, the expressions of the St curve and the LS curve are: (1) Where t represents time, S represents the longitudinal offset from the planning starting point in the Frenet coordinate system, L represents the lateral offset from the road centerline in the Frenet coordinate system, a1-a6 and b1-b6 represent the fifth-order polynomial coefficients of S with respect to t and the fifth-order polynomial coefficients of L with respect to S, respectively.

[0008] As a further illustration of the present invention, the longitudinal trajectory start and end position information is described as: (2) The starting and ending position information of the horizontal track is described as follows: (3) Where, 、 、 They represent the position, velocity and acceleration of the vehicle at the initial moment of collision avoidance respectively; 、 、 They represent the position, velocity and acceleration of the vehicle at the end of collision avoidance respectively; represents the road adhesion coefficient; is the vehicle width; Indicates the width of the obstacle vehicle ahead; Indicates the width of the road; The relative distance between the two vehicles on the central axis before the ego vehicle avoids the collision; It is the lateral redundant safety distance from the obstacle vehicle at the end of collision avoidance.

[0009] As a further illustration of the present invention, performing axial independent constrained optimization on the lateral and longitudinal candidate trajectory clusters specifically includes: Based on the road adhesion coefficient, the longitudinal acceleration and lateral acceleration constraint ranges are set in sequence; The trajectories whose longitudinal accelerations exceed the longitudinal acceleration constraint range in the longitudinal candidate trajectory cluster are eliminated, and the trajectories whose lateral accelerations exceed the lateral acceleration constraint range in the lateral candidate trajectory cluster are eliminated.

[0010] As a further illustration of the present invention, the longitudinal acceleration constraint range is , the lateral acceleration constraint range is .

[0011] As a further illustration of the present invention, the feasibility constraint optimization of the merged trajectory on the optimized lateral and longitudinal trajectory clusters specifically includes: Merge the optimized horizontal and vertical trajectory clusters; The trajectories that do not meet the preset safety, acceleration and yaw rate constraints in the merged trajectory cluster are eliminated.

[0012] As a further illustration of the present invention, the safety constraint condition is to ensure a certain preset safety distance between the ego vehicle and the obstacle vehicle, preferably 0.2m; the acceleration constraint condition is to constrain the combined ego vehicle acceleration based on the road adhesion coefficient. ; The yaw rate constraint is: ,in Indicates the maximum lateral acceleration allowed by the trajectory, preferably , is the trajectory velocity.

[0013] As a further illustration of the present invention, the optimal collision avoidance trajectory is selected from the trajectory clusters that meet the constraints through the objective function, specifically: An objective function is established based on evaluation indicators to sort the trajectories that meet the requirements and select the optimal collision avoidance trajectory. The evaluation indicators are lateral deviation, comfort, collision avoidance efficiency, and trajectory smoothness. The objective function is: (4) Wherein, the first term represents the lateral offset, i.e., the lateral position of the planned trajectory point; the second term represents the comfort, which takes into account the rate of change of lateral acceleration; the third term represents the collision avoidance time, which considers keeping the trajectory speed close to the speed at the start of collision avoidance; the fourth term represents the trajectory smoothness, which takes into account the curvature to reduce steering; 、 、 、 Respectively represent the weight coefficients corresponding to each indicator.

[0014] As a further illustration of the present invention, the longitudinal control uses a dual-loop PID controller to achieve closed-loop regulation of the speed-position error, specifically: The longitudinal tracking controller adopts a dual-loop PID control strategy, in which the first PID controller uses the deviation between the desired longitudinal position and the actual position of the vehicle to implement position closed-loop control; the second PID controller implements speed closed-loop control based on the speed error signal.

[0015] As a further illustration of the present invention, the position deviation and the speed deviation are expressed as: (5) Where, , Represents the actual longitudinal and lateral coordinates of the vehicle in the Cartesian coordinate system; , , Respectively represent the horizontal and vertical coordinates and reference heading angle of the planned trajectory point in the Cartesian coordinate system; is the velocity of the vehicle in the Cartesian coordinate system; is the heading angle of the vehicle in the Cartesian coordinate system, v is the heading angle of the vehicle projection point on the reference line; d and v represent the desired speed and actual speed of the vehicle respectively; is the curvature of the vehicle projection vector on the reference line; is the horizontal coordinate of the vehicle in the Frenet coordinate system.

[0016] As a further illustration of the present invention, an electric vehicle offline drive-brake system calibration table is established to directly convert the expected acceleration into the accelerator pedal opening or brake pressure to achieve longitudinal tracking.

[0017] Compared with the prior art, the present invention has the following beneficial technical effects: The present invention proposes a composite lateral and longitudinal collision avoidance control method for electric vehicles. Through hierarchical trajectory planning and decoupled tracking control, efficient coordination of braking and steering is achieved, overcoming the limitations of single braking or steering control, and improving the active safety performance of the vehicle under low-adhesion limit conditions.

[0018] Other features and advantages of this technical solution will be described in the subsequent description, and in part will become apparent from the description, or understood by practicing this technical solution. The objectives and other advantages of this technical solution can be achieved and obtained through the structures specifically pointed out in the written description and the accompanying drawings.

[0019] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings are used to provide a further understanding of the present technical solution and constitute a part of the specification. Together with the embodiments of the present technical solution, they are used to explain the present technical solution and do not constitute a limitation of the present technical solution. In the accompanying drawings: Figure 1 A schematic flow chart of a method for controlling a combined lateral and longitudinal collision avoidance of an electric vehicle provided by the present invention; Figure 2 Parameter conversion relationship diagram between the Cartesian coordinate system and the Frenet coordinate system constructed for the present invention; Figure 3 This is a schematic diagram of the safety distance constraint between the ego vehicle and the obstacle vehicle of the present invention; Figure 4 This is a schematic diagram of the longitudinal dual PID control of the present invention; Figure 5 It is a schematic diagram of the drive-brake calibration surface of the electric vehicle of the present invention.

[0021] Figure 6 Schematic diagram of the low adhesion coefficient 110km / h ~ 0.3 simulation working condition of the present invention, where (a)-(f) are the vehicle lateral position, lateral tracking error, longitudinal position, longitudinal tracking error, speed and heading angle change curves respectively.

[0022] Figure 7 This is a schematic diagram of collision avoidance in a simulated working condition with a low adhesion coefficient of 110 km / h ~ 0.3 according to the present invention, wherein: Figure 7 (a) is the trajectory of the right front point of the vehicle; Figure 7 (b) is the animation afterimage of the collision avoidance process when the ego vehicle overtakes the obstacle vehicle.

[0023] Figure 8 Schematic diagram of the low adhesion coefficient 120km / h ~ 0.3 simulation working condition of the present invention; (a)-(f) are the vehicle lateral position, lateral tracking error, longitudinal position, longitudinal tracking error, speed and heading angle change curves respectively.

[0024] Figure 9 This is a schematic diagram of the present invention's low adhesion coefficient 120km / h ~ 0.3 simulation working condition collision avoidance. Figure 9 (a) is the trajectory of the right front point of the vehicle; Figure 9 (b) is the animation afterimage of the collision avoidance process when the ego vehicle overtakes the obstacle vehicle. DETAILED DESCRIPTION

[0025] The preferred embodiments of the present technical solution are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present technical solution and are not used to limit the present technical solution.

[0026] With the increasing demand for electric vehicles to operate at high speeds and under complex road conditions, the application scenarios of vehicle active collision avoidance systems are gradually extending to emergency collision avoidance scenarios under low-adhesion conditions. In such scenarios, traditional single control strategies face severe challenges in two aspects: first, the risk of longitudinal braking failure: on low-adhesion roads, the friction coefficient between the tire and the ground is significantly reduced, and relying solely on longitudinal braking can easily lead to collision avoidance failure due to excessive braking distance; second, the risk of lateral lane change instability: under high-speed conditions, the vehicle's lateral dynamic characteristics are sensitive, and steering alone to avoid collisions may cause sideslip or yaw instability, especially under low-adhesion conditions, which increases the risk of loss of control. Most verification scenarios are based on high-adhesion roads or medium- and low-speed scenarios, lacking systematic simulation and evaluation under high-speed, low-adhesion extreme conditions.

[0027] To this end, the present invention proposes a composite lateral and longitudinal collision avoidance control method for electric vehicles, which achieves efficient coordination of braking and steering through hierarchical trajectory planning and decoupling tracking control, addresses the limitations of single braking or steering control, and improves the active safety performance of the vehicle under low-adhesion limit conditions.

[0028] Specifically, such as Figure 1 As shown, the present invention provides a composite collision avoidance control method suitable for electric vehicles under extreme operating conditions, comprising the following steps: Step 1: Based on the information of the ego vehicle, obstacle vehicles, and road, the lateral and longitudinal trajectory decoupling planning is performed in the Frenet coordinate system to obtain the lateral and longitudinal candidate trajectory clusters.

[0029] Specifically, the above step 1 includes the following process: Step 11: Convert the pose parameters of the ego vehicle and the obstacle vehicle from the Cartesian coordinate system to the Frenet coordinate system with the road centerline as the reference.

[0030] like Figure 2 As shown in the figure, a parameter conversion relationship diagram between the Cartesian coordinate system and the Frenet coordinate system is constructed. Based on the information of the ego vehicle, obstacle vehicle and road, the vehicle and obstacle posture parameters are converted into the Frenet coordinate system with the road centerline as the reference.

[0031] Step 12: Fit the lateral and longitudinal trajectories of the ego vehicle using a quintic polynomial to generate the LS and St curves, respectively. The St curve represents the longitudinal displacement of the ego vehicle described by a quintic polynomial with time t as the independent variable; the LS curve represents the lateral displacement described by a quintic polynomial with the longitudinal displacement S as the independent variable.

[0032] Among them, the expressions of St curve and LS curve are: (1) Where t represents time, S represents the longitudinal offset from the planning starting point in the Frenet coordinate system, L represents the lateral offset from the road centerline in the Frenet coordinate system, a1-a6 and b1-b6 represent the fifth-order polynomial coefficients of S with respect to t and the fifth-order polynomial coefficients of L with respect to S, respectively.

[0033] In this invention, longitudinal motion planning uses a quintic polynomial with time t as the independent variable to describe the longitudinal displacement; lateral motion planning uses a quintic polynomial with the longitudinal displacement S as the independent variable to describe the lateral displacement, representing the geometric offset relative to the reference path. This decoupling mechanism transforms the high-dimensional trajectory optimization problem into two low-degree-of-freedom optimization subproblems through dimensionality reduction, thereby improving the real-time performance of collision avoidance.

[0034] Step 13: Generate horizontal and vertical trajectory clusters according to the LS curve and St curve by setting the starting and ending position information of the horizontal and vertical trajectories.

[0035] The starting and ending position information of the longitudinal trajectory are described as: (2) The starting and ending position information of the horizontal track is described as: (3) Where, 、 、 They represent the position, velocity and acceleration of the vehicle at the initial moment of collision avoidance respectively; 、 、 They represent the position, velocity and acceleration of the vehicle at the end of collision avoidance respectively; represents the road adhesion coefficient; is the vehicle width; Indicates the width of the obstacle ahead. According to the national standard (GB 1589-2016), the width range is 1.6m~2.55m; Indicates the width of the road; The relative distance between the two vehicles on the central axis before the ego vehicle avoids the collision; The lateral redundant safety distance from the obstacle vehicle at the end of collision avoidance is generally 0.1m~0.3m, and the preferred distance in the present invention is 0.2m.

[0036] In the present invention, by setting the starting point and end point position information of the horizontal trajectory and the vertical trajectory planning, the horizontal and vertical trajectory clusters can be generated according to formula (1).

[0037] Step 2: The improved Lattice algorithm is used to perform axial independent constraint optimization on the lateral and longitudinal candidate trajectory clusters. The optimized lateral and longitudinal trajectory clusters are then subjected to combined trajectory feasibility constraint optimization. The trajectory clusters that meet the constraints are screened out as the optimal collision avoidance trajectory through the objective function.

[0038] The improved Lattice algorithm proposed in this paper prioritizes trackability by introducing a two-stage trajectory constraint mechanism. Specifically, in the first stage, independent axial constraints are applied before merging the lateral and longitudinal trajectories, avoiding wasted computational resources in the post-processing stage. In the second stage, feasibility constraints are applied after merging the lateral and longitudinal trajectories to ensure that the final output trajectory meets tracking requirements.

[0039] The above-mentioned axially independent constrained optimization of the lateral and longitudinal candidate trajectory clusters specifically includes the following process: Based on the road adhesion coefficient, the longitudinal acceleration and lateral acceleration constraint ranges are set in sequence. Preferably, the longitudinal acceleration constraint range is , the lateral acceleration constraint range is .

[0040] The trajectories whose longitudinal accelerations exceed the longitudinal acceleration constraint range in the longitudinal candidate trajectory cluster are eliminated, and the trajectories whose lateral accelerations exceed the lateral acceleration constraint range in the lateral candidate trajectory cluster are eliminated.

[0041] In the present invention, after creating the longitudinal and lateral trajectory clusters, the trajectories that do not meet the vehicle acceleration constraint conditions are directly eliminated. For the longitudinal trajectory cluster, once the longitudinal acceleration of a trajectory point exceeds range, the trajectory will be eliminated; for the lateral trajectory cluster, if the lateral acceleration of the trajectory point exceeds The track is then removed.

[0042] The feasibility constraint optimization of the merged trajectory for the optimized lateral and longitudinal trajectory clusters is performed as follows: The optimized lateral and longitudinal trajectory clusters are merged; the trajectories in the merged trajectory cluster that do not meet the preset safety, acceleration and yaw rate constraints are eliminated.

[0043] Safety constraint: Ensure that there is a certain preset safety distance between the vehicle and the obstacle vehicle, preferably 0.2m. Specifically, Figure 3 As shown in the figure, the ego vehicle and the obstacle vehicle are equivalent to rectangular frames. During the collision avoidance process, it is necessary to ensure that the ego vehicle rectangular frame and the obstacle vehicle rectangular frame are within a certain safe distance (0.2m in this paper); Acceleration constraints: Separate longitudinal and lateral constraints cannot guarantee that the combined acceleration is within a safe range. Therefore, the combined acceleration of the ego vehicle must still be constrained based on the road adhesion coefficient. ; Yaw rate constraint: To prevent the vehicle from losing control, the vehicle's yaw rate needs to be constrained. ,in Indicates the maximum lateral acceleration allowed by the trajectory, preferably , is the trajectory velocity.

[0044] In the present invention, after merging the longitudinal trajectory and the lateral trajectory after eliminating the ones that do not meet the acceleration constraint, the trajectory is still constrained for the second time. When the safety, acceleration and yaw angular velocity of the merged trajectory point do not meet the preset constraints, the trajectory is eliminated.

[0045] The above mentioned cluster of trajectories that meet the constraints is filtered out through the objective function to select the optimal collision avoidance trajectory. Specifically, the objective function is established based on the evaluation index to sort the trajectories that meet the requirements, and the optimal collision avoidance trajectory is filtered out and sent to the lower-level tracking controller for tracking. The evaluation index is the trajectory's lateral deviation, comfort, collision avoidance efficiency, and trajectory smoothness. The objective function established by the present invention is: (4) Wherein, the first term represents the lateral offset, i.e., the lateral position of the planned trajectory point. A large lateral offset in a short time at high speed will affect the stability. Under the premise of ensuring safety, the lateral offset should be as small as possible. The second term represents comfort, which mainly considers the rate of change of lateral acceleration. The third term represents the collision avoidance time, which mainly considers making the trajectory speed as close as possible to the speed at the start of collision avoidance to improve the collision avoidance efficiency. The fourth term represents the trajectory smoothness, which mainly considers the curvature. Excessive curvature will lead to abrupt turns. The smaller the curvature, the closer the trajectory is to a straight line, reducing the turning. 、 、 、 Respectively represent the weight coefficients corresponding to each indicator.

[0046] In view of the particularity of collision avoidance scenarios, the present invention takes into account the collision avoidance timeliness in addition to the lateral deviation, comfort and trajectory smoothness in order to improve the collision avoidance efficiency.

[0047] Step 3: Based on the optimal collision avoidance trajectory, a decoupled trajectory tracking controller is used to implement lateral and longitudinal composite collision avoidance control. The longitudinal control uses a dual-loop PID controller to achieve closed-loop adjustment of the speed-position error, and the lateral control uses an MPC tracker.

[0048] The above longitudinal control adopts a dual-loop PID controller to achieve closed-loop adjustment of speed-position error. Specifically, the longitudinal tracking controller adopts a dual-loop PID control strategy. The control diagram is shown in the figure below. Figure 4 As shown in the figure, the first PID controller uses the deviation between the desired longitudinal position and the actual position of the vehicle to implement position closed-loop control; the second PID controller implements speed closed-loop control based on the speed error signal.

[0049] Specifically, the position deviation and speed deviation can be expressed as: (5) Where, , Represents the actual longitudinal and lateral coordinates of the vehicle in the Cartesian coordinate system; , , Respectively represent the horizontal and vertical coordinates and reference heading angle of the planned trajectory point in the Cartesian coordinate system; is the velocity of the vehicle in the Cartesian coordinate system; is the heading angle of the vehicle in the Cartesian coordinate system, is the heading angle of the vehicle projection point on the reference line, v d and v represent the desired speed and actual speed of the vehicle respectively; is the curvature of the vehicle projection vector on the reference line; is the horizontal coordinate of the vehicle in the Frenet coordinate system.

[0050] In order to ensure real-time longitudinal control, an offline electric vehicle drive-brake system calibration table is established to directly convert the expected acceleration into the accelerator pedal opening or brake pressure to achieve longitudinal tracking. The specific electric vehicle drive-brake calibration surface diagram is as follows: Figure 5 As shown in Figure 2, the lateral control is tracked using an MPC controller.

[0051] Test scenario verification A joint simulation test system was built based on CarSim and MATLAB / Simulink. A complete vehicle model was created in CarSim based on a specific electric vehicle, and a high-speed, low-adhesion road extreme condition test simulation scenario was constructed to verify the safety and reliability of the collision avoidance system.

[0052] A snow-covered road was constructed in CarSim as a low-adhesion road simulation scenario, and the road adhesion coefficient was set to 0.3. The width of the obstacle vehicle was taken from the vehicle width and the maximum vehicle width specified in (GB 1589-2016). Finally, a simulation experiment was carried out based on the established simulation conditions.

[0053] ① Set the initial speed of the vehicle to 110 km / h. The distance between the vehicle and the stationary obstacle is 65 m, and the width of the obstacle is the same as that of the vehicle. Figure 6 The simulation results under this working condition are shown in Figure 2. The entire collision avoidance process lasts 2.65s, and the final lateral displacement of the vehicle is 2.3m. Figure 6 (b) and Figure 6 (d) It can be seen that the maximum lateral tracking error is about 0.014m, and the maximum longitudinal tracking error is about 0.082m, indicating that the trajectory tracker has a good tracking effect during the entire collision avoidance process. Figure 6 (e) The speed change curve shows that the vehicle speed gradually slows down from the initial 110km / h to about 98.6km / h. The maximum speed tracking error during the whole process is only 0.32km / h. Figure 6 (f) shows that the maximum heading tracking error of the ego vehicle is about 0.007 rad, but there is a certain lead phenomenon. However, the collision avoidance process pays more attention to position tracking, so the controller still shows good trajectory tracking performance.

[0054] The collision avoidance diagram of the vehicle under this working condition is as follows Figure 7 shown. Figure 7 (a) The trajectory of the right front point of the vehicle and Figure 7 (b) The afterimage of the collision avoidance animation of the ego vehicle overtaking the obstacle vehicle in CarSim shows that the ego vehicle did not come into contact with the obstacle vehicle during the entire collision avoidance process and maintained an effective safety distance.

[0055] ②Based on working condition ①, the vehicle speed is increased to 120km / h, the obstacle width is increased to 2.55m, and the distance between the two vehicles is 76m. The simulation results are as follows Figure 8 As shown in Figure 2, the collision avoidance process lasted 2.79 seconds, and the final lateral displacement of the vehicle was 2.75 meters. Figure 8 (b) shows that the maximum lateral tracking error is about 0.015m. Figure 8 (d) shows that the maximum longitudinal tracking error is 0.12m, which is slightly larger than that of working condition 1. This may be caused by the slightly larger speed tracking error during the collision avoidance process compared with working condition 1. In this working condition, the vehicle speed slows down from 120km / h to about 110.4km / h, and the maximum error is 0.76km / h. Figure 8 (f) The change in heading angle shows that the vehicle's heading angle is well tracked, with only slight fluctuations in the transition from collision avoidance to constant speed. Figure 9 As shown, the results show that the vehicle's driving trajectory maintains an effective safety distance from the obstacle vehicle during the vehicle collision avoidance process, and no contact occurs with the obstacle vehicle.

[0056] Simulation results show that under extreme high-speed, low-adhesion conditions, the collision avoidance system maintains a reasonable collision avoidance trajectory even when the width of the obstacle vehicle changes. The decoupled lateral and longitudinal trackers effectively track the collision avoidance trajectory. Furthermore, the vehicle maintains stable tracking during the tracking process, without losing control, skidding, or contacting the obstacle vehicle.

[0057] Obviously, those skilled in the art may make various changes and modifications to this technical solution without departing from the spirit and scope of this technical solution. Thus, if these modifications and variations of this technical solution fall within the scope of the claims of this technical solution and their equivalents, this technical solution is intended to include these modifications and variations.

Claims

1. A method for controlling the composite lateral and longitudinal collision avoidance of an electric vehicle, characterized in that: include: Based on the information of the ego vehicle, obstacle vehicles and road, the horizontal and vertical trajectory decoupling planning is performed in the Frenet coordinate system to obtain the horizontal and vertical candidate trajectory clusters; An improved Lattice algorithm is used to perform axial independent constraint optimization on the lateral and longitudinal candidate trajectory clusters. The optimized lateral and longitudinal trajectory clusters are then subjected to combined trajectory feasibility constraint optimization. The trajectory clusters that meet the constraints are screened out as the optimal collision avoidance trajectory through the objective function. Based on the optimal collision avoidance trajectory, a decoupled trajectory tracking controller is used to implement lateral and longitudinal composite collision avoidance control, wherein a dual-loop PID controller is used for longitudinal control to achieve closed-loop adjustment of speed-position error, and an MPC tracker is used for lateral control.

2. The method for controlling a combined lateral and longitudinal collision avoidance of an electric vehicle according to claim 1, wherein: Based on the information of the ego vehicle, the obstacle vehicle, and the road, the lateral and longitudinal trajectory decoupling planning is performed in the Frenet coordinate system to obtain the lateral and longitudinal candidate trajectory clusters, which specifically include: The pose parameters of the ego vehicle and the obstacle vehicle are converted from the Cartesian coordinate system to the Frenet coordinate system with the road centerline as the reference; The lateral and longitudinal trajectories of the ego vehicle are fitted with quintic polynomials to generate the LS and St curves, respectively. The St curve represents the longitudinal displacement of the ego vehicle described by a quintic polynomial with time t as the independent variable; the LS curve represents the lateral displacement described by a quintic polynomial with the longitudinal displacement S as the independent variable. By setting the starting point and end point position information of the horizontal track and the vertical track, and generating the horizontal and vertical track clusters according to the LS curve and the St curve.

3. The method for controlling a combined lateral and longitudinal collision avoidance of an electric vehicle as claimed in claim 2, wherein: The expressions of the St curve and the LS curve are: (1) Where t represents time, S represents the longitudinal offset from the planning starting point in the Frenet coordinate system, L represents the lateral offset from the road centerline in the Frenet coordinate system, a1-a6 and b1-b6 represent the fifth-order polynomial coefficients of S with respect to t and the fifth-order polynomial coefficients of L with respect to S, respectively.

4. The method for controlling a combined lateral and longitudinal collision avoidance of an electric vehicle as claimed in claim 2, wherein: The starting and ending position information of the longitudinal track is described as follows: (2) The starting and ending position information of the horizontal track is described as follows: (3) Where, 、 、 They represent the position, velocity and acceleration of the vehicle at the initial moment of collision avoidance respectively; 、 、 They represent the position, velocity and acceleration of the vehicle at the end of collision avoidance respectively; represents the road adhesion coefficient; is the vehicle width; Indicates the width of the obstacle vehicle ahead; Indicates the width of the road; The relative distance between the two vehicles on the central axis before the ego vehicle avoids the collision; It is the lateral redundant safety distance from the obstacle vehicle at the end of collision avoidance.

5. The method for controlling the combined lateral and longitudinal collision avoidance of an electric vehicle as claimed in claim 2, wherein: The performing axial independent constrained optimization on the lateral and longitudinal candidate trajectory clusters specifically includes: Based on the road adhesion coefficient, the longitudinal acceleration and lateral acceleration constraint ranges are set in sequence; The trajectories whose longitudinal accelerations exceed the longitudinal acceleration constraint range in the longitudinal candidate trajectory cluster are eliminated, and the trajectories whose lateral accelerations exceed the lateral acceleration constraint range in the lateral candidate trajectory cluster are eliminated.

6. The method for controlling the combined lateral and longitudinal collision avoidance of an electric vehicle as claimed in claim 2, wherein: The feasibility constraint optimization of the merged trajectory for the optimized lateral and longitudinal trajectory clusters specifically includes: Merge the optimized horizontal and vertical trajectory clusters; The trajectories that do not meet the preset safety, acceleration and yaw rate constraints in the merged trajectory cluster are eliminated.

7. The method for controlling a combined lateral and longitudinal collision avoidance of an electric vehicle as claimed in claim 1, wherein: The optimal collision avoidance trajectory is selected from the trajectory clusters that meet the constraints through the objective function, specifically: An objective function is established based on evaluation indicators to sort the trajectories that meet the requirements and select the optimal collision avoidance trajectory. The evaluation indicators are lateral deviation, comfort, collision avoidance efficiency, and trajectory smoothness. The objective function is: (4) Wherein, the first term represents the lateral offset, i.e., the lateral position of the planned trajectory point; the second term represents the comfort, which takes into account the rate of change of lateral acceleration; the third term represents the collision avoidance time, which considers keeping the trajectory speed close to the speed at the start of collision avoidance; the fourth term represents the trajectory smoothness, which takes into account the curvature to reduce steering; 、 、 、 Respectively represent the weight coefficients corresponding to each indicator.

8. The method for controlling the combined lateral and longitudinal collision avoidance of an electric vehicle as claimed in claim 1, wherein: The longitudinal control adopts a dual-loop PID controller to achieve closed-loop adjustment of speed-position error, specifically: The longitudinal tracking controller adopts a dual-loop PID control strategy, in which the first PID controller uses the deviation between the desired longitudinal position and the actual position of the vehicle to implement position closed-loop control; the second PID controller implements speed closed-loop control based on the speed error signal.

9. The method for controlling the combined lateral and longitudinal collision avoidance of an electric vehicle as claimed in claim 8, characterized in that: The position error and speed error are expressed as: (5) Where, , Represents the actual longitudinal and lateral coordinates of the vehicle in the Cartesian coordinate system; , , Respectively represent the horizontal and vertical coordinates and reference heading angle of the planned trajectory point in the Cartesian coordinate system; is the velocity of the vehicle in the Cartesian coordinate system; is the heading angle of the vehicle in the Cartesian coordinate system, v is the heading angle of the vehicle projection point on the reference line; d and v represent the desired speed and actual speed of the vehicle respectively; is the curvature of the vehicle projection vector on the reference line; is the horizontal coordinate of the vehicle in the Frenet coordinate system.

10. The electric vehicle's lateral and longitudinal combined collision avoidance control method as claimed in claim 8, characterized in that: By establishing an offline drive-brake system calibration table for electric vehicles, the expected acceleration is directly converted into accelerator pedal opening or brake pressure to achieve longitudinal tracking.

Images