Multi-constraint parking path planning method, device and automobile control system

Abstract

The present application relates to a multi-constraint parking path planning method, device and automobile control system, wherein the multi-constraint parking path planning method includes: determining multiple candidate parking start positions and parking end positions of the vehicle based on the dung beetle optimization algorithm and the current position of the vehicle, the path of the vehicle moving from the current position to the parking start position is the first path, and the path of the vehicle moving from the current position to the parking end position is the second path; determining the target parking start position and target parking end position of the vehicle and the corresponding first path and second path from the multiple candidate parking start positions and parking end positions of the vehicle through the target optimization model. The present invention mainly divides the parking process into two stages, the first stage is used to adjust the vehicle position and the second stage is used to perform the parking action, and the path end points of the two stages are directly calculated by the improved dung beetle optimization algorithm, thereby improving the path planning efficiency of automatic parking.

Description

Technical Field

[0001] The present application relates to the field of autonomous driving, and in particular to a multi-constraint parking path planning method, device and vehicle control system. Background Art

[0002] The types of parking path planning for driverless cars include polynomial curves, arc-straight lines, B-spline curves, sigmoid functions, and arc tangent function curve planning. For example, the five-segment parallel parking path planning of straight line-transition curve-arc segment-transition curve-arc segment meets the curvature continuity requirement and has low path tracking difficulty; the boundary of the feasible parking starting area for parallel parking is derived based on the arc-straight line combination curve, and the curvature-continuous parallel parking path is obtained by rotating and flipping the fixed clothoid curve.

[0003] The above-mentioned parking path planning methods for driverless cars are mostly single-step parking methods, that is, the current position of the car is directly used as the starting point to plan a parking route into the parking space. However, the current position of the car is determined by the driver (usually the driver starts the automatic parking function after parking), and thus the current position of the car is not an ideal parking starting point. Therefore, the above-mentioned parking path planning methods for driverless cars cannot plan a better parking path in some cases, or even cannot plan a parking path, and a poor parking path will make the final position of the vehicle unsatisfactory or cause a collision with an obstacle. Summary of the invention

[0004] The present invention provides a multi-constraint parking path planning method, device and vehicle control system to solve the problem that the current parking path planning method for unmanned vehicles cannot plan a better parking path or even cannot plan a parking path.

[0005] In a first aspect, the present invention provides a multi-constraint parking path planning method, comprising:

[0006] Determine multiple candidate parking start positions and parking end positions of the vehicle based on the dung beetle optimization algorithm and the current position of the vehicle, the path of the vehicle moving from the current position to the parking start position is the first path, and the path of the vehicle moving from the current position to the parking end position is the second path;

[0007] Determining a target parking start position and a target parking end position of the vehicle and a corresponding first path and a second path from a plurality of candidate parking start positions and parking end positions of the vehicle through a target optimization model;

[0008] The target optimization model includes an objective function and optimization constraints. The objective function takes minimizing the total length of the first path and the second path as the optimization goal. The optimization constraints include a collision constraint between the first path and the second path and a curvature constraint of the second path.

[0009] In a second aspect, the present invention provides a multi-constraint parking path planning device, comprising:

[0010] A position optimization module, for determining a plurality of candidate parking start positions and parking end positions of the vehicle based on the dung beetle optimization algorithm and the current position of the vehicle, wherein the path of the vehicle moving from the current position to the parking start position is the first path, and the path of the vehicle moving from the current position to the parking end position is the second path;

[0011] A path determination module, used to determine a target parking start position and a target parking end position of the vehicle and a corresponding first path and a second path from a plurality of candidate parking start positions and parking end positions of the vehicle through a target optimization model;

[0012] The target optimization model includes an objective function and optimization constraints. The objective function takes minimizing the total length of the first path and the second path as the optimization goal. The optimization constraints include a collision constraint between the first path and the second path and a curvature constraint of the second path.

[0013] In a third aspect, the present invention provides a vehicle control system, comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the multi-constraint parking path planning method described in the first aspect.

[0014] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the multi-constraint parking path planning method described in the first aspect are implemented.

[0015] Compared with the related art, the present invention mainly divides the parking process into two stages. The first stage is used to adjust the vehicle position, that is, to adjust the parking starting position, and the parking action is not performed in this stage. The second stage is used to perform the parking action, that is, to control the vehicle to gradually enter the parking space. Even if the current position of the vehicle is not good, the adjustment in the first stage can also make the vehicle in a better parking starting position, and then through the second stage, the vehicle can enter the parking space better. In this process, the final posture of the vehicle can be made more ideal, such as being in the center of the parking space and parallel to the parking space, so that the subsequent driver can drive the vehicle out. At the same time, the probability of collision with obstacles during parking in the parking space can be greatly reduced. Further, the operating difficulty of the driver is also reduced, that is, the driver does not need to park the vehicle at the ideal position before starting the automatic parking function. Therefore, the present invention can adapt to more complex parking road conditions and improve the planning quality of the parking path. In addition, the path end points of the two stages are directly calculated by the improved dung beetle optimization algorithm, and no other calculation process is required, which also improves the path planning efficiency of automatic parking. Therefore, the present invention solves the problem that the current parking path planning method for an unmanned vehicle cannot plan a better parking path or even cannot plan a parking path.

[0016] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a flow chart of a multi-constraint parking path planning method provided in an embodiment of the present invention;

[0018] Figure 2 is a schematic diagram of the parallel parking process in the present invention;

[0019] Figure 3 is a schematic diagram of the process of vertical parking in the present invention;

[0020] Figure 4 It is a schematic diagram of the range of the parking start position and the parking end position during parallel parking in the present invention;

[0021] Figure 5 It is a schematic diagram of the range of the parking start position and the parking end position during vertical parking in the present invention;

[0022] Figure 6 It is a schematic diagram of possible collision situations in parallel parking;

[0023] Figure 7 It is a schematic diagram of possible collision situations in perpendicular parking;

[0024] Figure 8This is a comparison chart of iterative tests of different algorithms on the standard test function F4;

[0025] Fig. 9 This is a comparison chart of iterative tests of different algorithms on the standard test function F7;

[0026] Fig.10 This is a comparison chart of iterative tests of different algorithms on the standard test function F8;

[0027] Fig.11 This is a comparison chart of iterative tests of different algorithms on the standard test function F9;

[0028] Fig.12 This is a comparison chart of iterative tests of different algorithms on the standard test function F10;

[0029] Fig.13 This is a comparison chart of iterative tests of different algorithms on the standard test function F14;

[0030] Fig.14 This is a comparison chart of iterative tests of different algorithms on the standard test function F20;

[0031] Fig.15 This is a comparison chart of iterative tests of different algorithms on the standard test function F23;

[0032] Fig.16 It is the comparison result of iterative calculation of DBO and IDBO for parallel parking in the present invention;

[0033] Fig.17 is an optimal path diagram for parallel parking provided by a multi-constraint parking path planning method in one embodiment of the present invention;

[0034] Fig.18 yes Fig.17 The first path curvature diagram in the parallel parking optimal path diagram;

[0035] Fig.19 yes Fig.17 The curvature diagram of the second path in the optimal path diagram for parallel parking;

[0036] Fig. 20 The iterative calculation comparison results of DBO and IDBO for vertical parking in the present invention are

[0037] Fig.21 is an optimal path diagram for vertical parking provided by a multi-constraint parking path planning method in one embodiment of the present invention;

[0038] Fig. 22 yes Fig.21 The first path curvature diagram in the vertical parking optimal path diagram;

[0039] Fig.23 yes Fig.21 The second path curvature diagram in the vertical parking optimal path diagram;

[0040] Fig.24 The figure is a comparison between the simulated path and the theoretical path for parallel parking;

[0041] Fig.25 A comparison diagram of the simulated heading angle and the theoretical heading angle for the first path in parallel parking;

[0042] Fig.26 A comparison diagram of the simulated heading angle and the theoretical heading angle of the second path in parallel parking;

[0043] Fig. 27 The comparison diagram of the simulated path and the theoretical path for vertical parking;

[0044] Fig.28 A comparison diagram of the simulated heading angle and the theoretical heading angle for the first path in perpendicular parking;

[0045] Fig.29 The figure is a comparison diagram between the simulated heading angle and the theoretical heading angle of the second path in perpendicular parking. DETAILED DESCRIPTION

[0046] In order to more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0047] Unless otherwise defined, the technical terms or scientific terms involved in this application shall have the general meaning understood by people with general skills in the technical field to which this application belongs. The words "one", "a", "the", "these" and the like in this application do not indicate a quantitative limitation, and they may be singular or plural. The terms "include", "comprise", "have" and any variants thereof involved in this application are intended to cover non-exclusive inclusions; for example, a process, method and system, product or device comprising a series of steps or modules (units) is not limited to the listed steps or modules (units), but may include unlisted steps or modules (units), or may include other steps or modules (units) inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there may be three relationships, for example, "A and / or B" may mean: A exists alone, A and B exist at the same time, and B exists alone. Generally, the character " / " indicates that the objects associated with each other are in an "or" relationship. The terms "first", "second", "third", etc. in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0048] In an embodiment of the present invention, a multi-constraint parking path planning method is provided. Figure 1 is a flowchart of a multi-constraint parking path planning method provided in an embodiment of the present invention. Figure 1 As shown, the process includes step S110 and step S120.

[0049] Step S110, based on the dung beetle optimization algorithm and the current position of the vehicle, multiple candidate parking start positions and parking end positions of the vehicle are determined, the path of the vehicle moving from the current position to the parking start position is the first path, and the path of the vehicle moving from the current position to the parking end position is the second path.

[0050] It should be noted that the positions referred to in the present invention are all point positions. For the vehicle, a reference point in the vehicle can be selected as the vehicle position point, and the path is the moving trajectory of the vehicle position point. Exemplarily, the reference point can be the center of the rear axle of the vehicle. The current position of the vehicle is the position when the method is executed, that is, the initial position of the vehicle during path planning.

[0051] Figure 2 is a schematic diagram of the parallel parking process in the present invention, Figure 3 Schematic diagram of the vertical parking process in the present invention. Figure 2 and Figure 3 . In the present invention, the parking process is divided into two stages. The first stage is used to adjust the position of the vehicle, that is, to adjust the parking starting position, and the parking action is not performed in this stage. The second stage is used to perform the parking action, that is, to control the vehicle to gradually enter the parking space. Specifically, if the parking space is on the side of the road where the vehicle is located, the first stage is to adjust the position of the vehicle on the road, and the second stage is to control the vehicle to enter the parking space from the road. The entire parking process is divided into two stages. Even if the current position of the vehicle is not good, the adjustment of the first stage can make the vehicle in a better parking starting position, and then through the second stage, the vehicle can enter the parking space better. In this process, the final position of the vehicle can be made more ideal, such as being in the center of the parking space and parallel to the parking space, so that the subsequent driver can drive the vehicle out. At the same time, the probability of collision with obstacles during parking can be greatly reduced. Further, the operating difficulty of the driver is also reduced, that is, the driver does not need to park the vehicle in the ideal position before starting the automatic parking function.

[0052] At the same time, refer to Figure 2 and Figure 3 It can be seen that parallel parking refers to parking when the parking position is parallel to the lane, and perpendicular parking refers to parking when the parking position is perpendicular to the lane.

[0053] Furthermore, in order to determine the first path and the second path during the parking process, it is first necessary to determine the end point of the first path (the starting point of the second path) and the end point of the second path. After the starting point (the current position of the vehicle) and the end point of the first path are determined, the first path can be further determined, and after the starting point and the end point of the second path are determined, the second path can be further determined.

[0054] Specifically, for the starting point and end point of the two paths, the present invention applies the dung beetle optimization algorithm to determine them. By continuously iteratively updating the current position of the vehicle through the dung beetle optimization algorithm, the starting point and end point of the two paths can be determined. The dung beetle optimizer (DBO) is a swarm intelligence optimization algorithm proposed by Jiankai Xue et al. in 2022. It is inspired by the biological behavior process of dung beetles and has the characteristics of strong optimization ability and fast convergence speed. The inspiration of this algorithm mainly comes from the rolling, dancing, foraging, stealing, and reproduction behaviors of dung beetles. The traditional dung beetle optimization algorithm is a mature technology and is not specifically introduced in the present invention. Taking into account that the traditional dung beetle optimization algorithm has certain limitations in parking problems (it can also be used), the present invention also improves the traditional dung beetle optimization algorithm so that the dung beetle optimization algorithm can be better applied in the present invention.

[0055] Accordingly, in a preferred embodiment, the dung beetle optimization algorithm is an improved dung beetle optimization algorithm, and the improved dung beetle optimization algorithm includes an improved rolling ball dung beetle strategy, an egg ball breeding strategy and a foraging dung beetle strategy. Step S110 specifically includes step S111, step S112 and step S113.

[0056] Step S111: Based on the current position of the vehicle, at least one candidate parking start position and parking end position of the vehicle is determined by using an improved rolling ball strategy.

[0057] The original rolling ball dung beetle strategy only considers the optimal value of the population in the previous iteration as a reference, but does not consider the historical optimal value of each dung beetle itself as a reference, which will increase the number of iterations of the algorithm as a whole. Therefore, the rolling ball dung beetle strategy is improved.

[0058] Improved rolling ball dung beetle strategies include:

[0059]

[0060] in, r 1. r 2. n 1 and n 2 represent different parameters respectively. a 1 and a 2 represents a random number between 0 and 1, t represents the number of iterations, X i ( t ) indicates the t The iteration i The location of the dung beetle, X i ( t+ 1) Indicates the t +1 iteration i The location of the dung beetle, bestX It represents the optimal individual position in the current dung beetle population. Dung beetles are used to represent the vehicle.

[0061] Step S112: Based on the current position of the vehicle, at least one candidate parking start position and parking end position of the vehicle is determined by using an improved egg ball breeding strategy.

[0062] Since all dung beetles in the original egg ball breeding strategy move in a straight line with reference to the current optimal solution of the population, it is very likely that the population will fall into a local optimal solution, weakening the global search ability of the algorithm in the early stage. Therefore, the egg ball breeding strategy is improved.

[0063] Improved egg ball propagation strategies include:

[0064]

[0065] in, t represents the number of iterations, X i ( t ) indicates the t The iteration i The location of the dung beetle, X i ( t+ 1) Indicates the t +1 iteration i The location of the dung beetle, bestX represents the optimal individual position in the current dung beetle population. Dung beetles are used to represent the vehicle. r Indicates the parameters, g is a constant, M represents the maximum number of iterations, Represents a random number between -1 and 1. and They respectively represent the lower and upper limits of the dung beetle egg-laying area (the upper and lower limits of the position point coordinates). The dung beetle egg-laying area represents the area where the vehicle can stay.

[0066] Step S113: Based on the current position of the vehicle, at least one candidate parking start position and parking end position of the vehicle is determined by using an improved foraging dung beetle strategy.

[0067] The problem of the original foraging dung beetle strategy is similar to that of the original rolling ball dung beetle strategy. It only considers the optimal value of the population in the previous iteration as a reference without considering the historical optimal value of each dung beetle itself. In order to further reduce the overall number of iterations of the algorithm, dynamic weight adjustment is added to prevent the early convergence from being too fast, and the original foraging dung beetle strategy is improved.

[0068] Improved foraging strategies for dung beetles include:

[0069]

[0070] in, t represents the number of iterations, X i ( t ) indicates the t The iteration i The location of the dung beetle, X i ( t+ 1) Indicates the t +1 iteration i The location of the dung beetle, bestX represents the optimal individual position in the current dung beetle population. Dung beetles are used to represent the vehicle. C 1 represents a random number that follows a normal distribution. C 2 represents a random vector belonging to (0, 1), Lb b and Ub b They represent the lower and upper limits of the dung beetle foraging area (the upper and lower limits of the position point coordinates), respectively. The dung beetle foraging area represents the area where the vehicle can stay. M Indicates the maximum number of iterations.

[0071] Step S110 may further include step S114, in which at least one candidate parking start position and parking end position of the vehicle is determined by a stealing strategy based on the current position of the vehicle. The original dung beetle optimization algorithm also includes a stealing strategy, so the stealing strategy may also be used to simultaneously determine the parking start position and parking end position of the vehicle. It should be noted that the present invention does not improve the stealing strategy, and the stealing strategy in the original dung beetle optimization algorithm may be directly used.

[0072] In the above process, different strategies in the improved dung beetle optimization algorithm are used to optimize the parking start position and the parking end position of the vehicle at the same time. Compared with using only one strategy, it is possible to avoid falling into a local optimal solution in the optimization problem of step S120. The dung beetle population of each strategy includes multiple dung beetles, which represent different possible positions of the vehicle, that is, multiple pairs of parking start positions and parking end positions of the vehicle (corresponding to the number of dung beetles) can be obtained through the improved dung beetle optimization algorithm.

[0073] Among them, in the dung beetle optimization algorithm, X Indicates the position of the dung beetle, that is, the position of the vehicle. Since the present invention needs to determine the parking start position and the parking end position, it can be X It is formed in a four-dimensional vector form, which contains the abscissa and ordinate of the parking start position and the abscissa and ordinate of the parking end position. X The initial value of is the current position of the vehicle.

[0074] In both the improved egg ball breeding strategy and the foraging dung beetle strategy, it is necessary to determine the area where the vehicle can stay, that is, to limit the range of the parking start position and the parking end position.

[0075] Figure 4 is a schematic diagram of the range of the parking start position and the parking end position during parallel parking in the present invention, Figure 5 It is a schematic diagram of the range of the parking start position and the parking end position during vertical parking in the present invention.

[0076] Reference Figure 4 and Figure 5, the range of the parking starting position (area 1 in the figure) is mainly determined according to the lane width, and the main thing is to ensure that the vehicle does not collide with the lane boundary. When the rear axle center of the vehicle is used as the vehicle position point, if the vehicle position point is at the upper boundary of the range, the upper boundary of the vehicle is just aligned with the upper boundary of the lane, and if the vehicle position point is at the lower boundary of the range, the lower boundary of the vehicle is just aligned with the lower boundary of the lane.

[0077] Reference Figure 4 and Figure 5 , the range of the parking end position (area 2 in the figure) is mainly determined according to the size of the parking space, and the main thing is to ensure that the vehicle does not collide with the parking space boundary. When the rear axle center of the vehicle is used as the vehicle position point, if the vehicle position point is at the upper boundary of the range, the upper boundary of the vehicle is just aligned with the upper boundary of the parking space, if the vehicle position point is at the left boundary of the range, the left boundary of the vehicle is just aligned with the left boundary of the parking space, if the vehicle position point is at the lower boundary of the range, the lower boundary of the vehicle is just aligned with the lower boundary of the parking space, and if the vehicle position point is at the right boundary of the range, the right boundary of the vehicle is just aligned with the right boundary of the parking space.

[0078] Step S120, determining a target parking start position and a target parking end position of the vehicle and the corresponding first path and second path from a plurality of candidate parking start positions and parking end positions of the vehicle through a target optimization model; wherein the target optimization model includes an objective function and optimization constraints, the objective function takes minimizing the total length of the first path and the second path as an optimization goal, and the optimization constraints include collision constraints of the first path and the second path and curvature constraints of the second path.

[0079] Specifically, the objective function can be set as:

[0080]

[0081] in, d 1 and d 2 represents the total number of sampling points of the first path and the second path respectively, x i Indicates the first path i The horizontal coordinate of the sampling point, x i-1 Indicates the first path i -1 The horizontal coordinate of the sampling point, F 1 represents the first path function, x j Indicates the second path j The horizontal coordinate of the sampling point, x j-1 Indicates the second path j -1 The horizontal coordinate of the sampling point, F 2 represents the second path function.

[0082] The above objective function can calculate the total length of each pair of the first path and the second path. In order to simplify the calculation difficulty, the paths are sampled and the distances between each adjacent position point are accumulated. The first path function is used to characterize the relationship between the horizontal coordinate and the vertical coordinate of each position point in the first path, and the second path function is used to characterize the relationship between the horizontal coordinate and the vertical coordinate of each position point in the second path. Therefore, F 1( x i ) indicates the first path i The vertical coordinate of the sampling point, F 1( x i-1 ) indicates the first path i- The vertical coordinate of a sampling point, F 2( x j ) indicates the first j The vertical coordinate of the sampling point, F 2( x j-1 ) indicates the first j -1 The vertical coordinate of the sampling point.

[0083] Specifically, the first path function is:

[0084]

[0085]

[0086] in, a 0. a 1. a 2. a 3. a 4 and a 5 represent different coefficients, x 0 and y 0 represents the horizontal and vertical coordinates of the current position of the vehicle, x a and y a They represent the horizontal and vertical coordinates of the parking starting position of the vehicle, θ Indicates the current heading angle of the vehicle (the initial heading angle of the vehicle during path planning, that is, the heading angle when the driver parks the vehicle, which can be obtained by measurement).

[0087] It can be seen from the above function that when the starting point and the end point of the first path and the current heading angle of the vehicle are determined, the specific values ​​of the six coefficients in the first path function can be calculated, and then the specific expression of the first path function can be obtained.

[0088] Specifically, the second path function is:

[0089]

[0090] In parallel parking maneuvers:

[0091]

[0092] In a perpendicular parking maneuver:

[0093]

[0094] in, b 0. b 1. b 2. b 3. b 4 and b 5 represent different coefficients, x b and y b They respectively represent the horizontal and vertical coordinates of the parking end position of the vehicle.

[0095] Since the second paths in parallel parking and perpendicular parking are different, the specific expressions of the second path function in the two conditions are different. It can be seen from the above function that when the starting point and the end point of the second path are determined, the specific values ​​of the coefficients in the second path function can be calculated, and then the specific expression of the second path function can be obtained.

[0096] It should be noted that for each strategy, after completing all iterations, the best dung beetle is selected as the optimal solution of the strategy. Then, the global optimal dung beetle is selected from the optimal dung beetles of different strategies, and the parking start position and parking end position corresponding to the global optimal dung beetle are used as the target parking start position and target parking end position, that is, the final parking start position and parking end position, and the corresponding first path and second path are used as the final parking path.

[0097] In general, the present invention mainly divides the parking process into two stages, and obtains different parking start positions and parking end positions through the improved dung beetle optimization algorithm. Each pair of parking start positions and parking end positions corresponds to a pair of first paths and second paths. The above-mentioned objective function calculates the total length of each pair of first paths and second paths, and takes the parking start position and parking end position corresponding to the first path and second path with the smallest total length as the target parking start position and target parking end position, so as to obtain the parking path with the highest parking efficiency. The target parking start position and target parking end position are the parking start position and parking end position finally selected, and the corresponding first path and second path are the parking path finally selected. The optimization target of the improved dung beetle optimization algorithm is to update the parking start position and parking end position that make the objective function value smaller, that is, the objective function is the fitness function of the improved dung beetle optimization algorithm. Among them, it is more critical that the parking process is divided into two stages. The first stage is used to adjust the vehicle position, that is, to adjust the parking start position, and the parking action is not performed in this stage. The second stage is used to perform the parking action, that is, to control the vehicle to gradually enter the parking space. Even if the current position of the vehicle is not good, the adjustment in the first stage can make the vehicle in a better parking starting position, and then through the second stage, the vehicle can enter the parking space better. In this process, the final posture of the vehicle can be made more ideal, such as being in the center of the parking space and parallel to the parking space, so that the subsequent driver can drive the vehicle out. At the same time, the probability of collision with obstacles during parking can be greatly reduced. Further, the driver's operating difficulty is also reduced, that is, the driver does not need to park the vehicle in the ideal position before starting the automatic parking function. Therefore, the present invention can adapt to more complex parking conditions and improve the planning quality of the parking path. In addition, the path end points of the two stages are directly calculated by the improved dung beetle optimization algorithm, and no other calculation process is required, which also improves the path planning efficiency of automatic parking. Therefore, the present invention solves the problem that the current unmanned vehicle parking path planning method cannot plan a better parking path or even a parking path.

[0098] In this process, the selected path should satisfy some constraints, and the specific constraints can be set according to the specific situation, such as the collision constraint between the first path and the second path and the curvature constraint of the second path. Among them, the curvature constraint is used to ensure that the turning radius of the ego vehicle is greater than the minimum turning radius, that is, the curvature radius corresponding to each position point on the second path is greater than the minimum turning radius. The minimum turning radius is determined by the ego vehicle, that is, the turning radius when the vehicle turns the largest angle. The collision constraint includes the collision constraint of the first path and the collision constraint of the second path. Figure 6 This is a schematic diagram of possible collision situations in parallel parking. Figure 7 is a schematic diagram of possible collision situations in perpendicular parking. Figure 6 and Figure 7 , taking the parking space below the lane where the ego vehicle is located as a reference: the collision constraint of the first path is used to ensure that the ego vehicle does not collide with the upper and lower boundaries of the lane; in the parallel parking condition, the collision constraint of the second path is used to ensure that the ego vehicle does not collide with the upper right corner and the lower boundary of the parking space; in the perpendicular parking condition, the collision constraint of the second path is used to ensure that the ego vehicle does not collide with the upper right corner and the left boundary of the parking space.

[0099] The above is a relatively complete introduction to the multi-constraint parking path planning method provided by the present invention. There are two key points: one is to divide the parking process into two stages, and the other is to apply and improve the dung beetle optimization algorithm so that it can better calculate the terminal positions of the two parking stages. The effectiveness of the improved dung beetle optimization algorithm is demonstrated as follows.

[0100] The present invention uses eight typical standard test functions F4, F7, F8, F9, F10, F14, F20, and F23 in the CEC standard test set (a total of 23 test functions) for test comparison. Four swarm intelligence algorithms, GWO (Gray Wolf Optimization Algorithm), SSA (Sparrow Optimization Algorithm), DBO (Dung Beetle Optimization Algorithm) and IDBO (the improved dung beetle optimization algorithm of the present invention), are selected for comparison, and the iterative effect comparison of each standard test function is given. Among them:

[0101]

[0102] in, D To solve for the dimension, f min The above functions are all existing mature functions, and the present invention will not provide specific explanations.

[0103] Figure 8 This is a comparison chart of iterative tests of different algorithms on the standard test function F4; Fig. 9 This is a comparison chart of iterative tests of different algorithms on the standard test function F7; Fig.10 This is a comparison chart of iterative tests of different algorithms on the standard test function F8; Fig.11 This is a comparison chart of iterative tests of different algorithms on the standard test function F9; Fig.12 This is a comparison chart of iterative tests of different algorithms on the standard test function F10; Fig.13 This is a comparison chart of iterative tests of different algorithms on the standard test function F14; Fig.14 This is a comparison chart of iterative tests of different algorithms on the standard test function F20; Fig.15 This is a comparison chart of iterative tests of different algorithms on the standard test function F23.

[0104] Through the above comparison, it can be seen that among the four algorithms, for any one of the eight test functions, the dung beetle optimization algorithm improved by the present invention can quickly iterate and converge to the optimal value, and can also quickly jump out of the local optimal solution during the iteration process. This can reflect a greater advantage than the unimproved dung beetle optimization algorithm, and at the same time improves the accuracy of the final optimal solution.

[0105] We still choose four swarm intelligence algorithms, namely GWO (Gray Wolf Optimization Algorithm), SSA (Sparrow Optimization Algorithm), DBO (Dung Beetle Optimization Algorithm), and IDBO (Dung Beetle Optimization Algorithm improved in this paper) for comparison. Each algorithm runs each standard test function fifty times, and the average value, standard deviation and optimal value of these fifty times are calculated. The test results are shown in the following table.

[0106] Table 1 The running results of each algorithm on different standard test functions

[0107]

[0108] Through the above comparison, it can be seen that the improved dung beetle optimization algorithm of the present invention can still stably reflect the advantages of the improvement under a large number of tests. In eight test functions of four categories, the accuracy of the optimal value solved is stably improved compared with the algorithm before improvement, and the standard deviation and average value of the optimal value solved by the improved algorithm are the best among the four algorithms. The above comparative test better reflects the effectiveness of the improvement strategy used in the present invention.

[0109] As follows, the implementation process of the multi-constraint parking path planning method provided by the present invention is described through a specific example.

[0110] 1. Parallel parking.

[0111] The center of the rear axle of the vehicle is the vehicle position point. The initial position of the vehicle is set to (-5, 3), the initial heading angle is set to 0, the origin of the coordinate system is set to the upper left corner of the parking space, and the specific parameters of the vehicle are as shown in the following table.

[0112] Table 2 Vehicle parameters

[0113]

[0114] Referring to the dimensions recommended by the national standard, the length of the parallel parking space is set to 6.5m and the width is set to 2.5m. Considering the unstructured road, the road width is set to 6m. Input the above parameters and use the improved dung beetle optimization algorithm to solve the path and the coordinates of the relevant path endpoints. At the same time, the traditional dung beetle optimization algorithm (DBO) is used to perform iterative calculations of the path for comparison. Fig.16 It is the comparison result of iterative calculation of DBO and IDBO for parallel parking in the present invention.

[0115] Reference Fig.16 , the optimal path length obtained by IDBO is 20.1411, x a =7.6400, y a =1.2813, x b =0.7500, y b =-0.8575. Specific path reference Fig.17 .

[0116] Depend on Fig.17 It can be seen that the vehicle has no collision with the road and the parking space boundary, satisfying the collision constraint.

[0117] The curvatures of the first and second paths in the optimal path are respectively Fig.18 and Fig.19 It can be seen that the absolute value of the curvature of the first path and the second path does not exceed 0.25, which satisfies the constraint of the minimum turning radius of the vehicle.

[0118] 1. Perpendicular parking.

[0119] The initial position of the vehicle, the initial heading angle of the vehicle and the specific parameters of the vehicle are the same as those of parallel parking. Referring to the recommended dimensions of the national standard, the length of the vertical parking space is set to 5.3m and the width is set to 2.5m. Input the above parameters, use the improved dung beetle optimization algorithm to solve the path and the coordinates of the relevant path endpoints, and compare them with the DBO algorithm iteration. At the same time, the traditional dung beetle optimization algorithm (DBO) is used to perform iterative calculations of the path for comparison. Fig. 20 It is the comparison result of iterative calculation of DBO and IDBO for vertical parking in the present invention.

[0120] Reference Fig. 20 , the optimal path length obtained by IDBO is 27.8914, x a =9.9128, y a =3.2145, x b =0.9550, y b =-3.2000. Specific path reference Fig.21 .

[0121] Reference Fig.21 It can be seen that the vehicle has no collision with the road and the parking space boundary, satisfying the collision constraint.

[0122] The curvatures of the first and second paths in the optimal path are respectively Fig. 22 and Fig.23It can be seen that the absolute value of the curvature of the first path and the second path does not exceed 0.25, which satisfies the constraint of the minimum turning radius of the vehicle.

[0123] The above are the theoretical calculation results of path planning. We will continue to simulate the path tracking control effect under the same scenario settings.

[0124] Use matlab-carsim-prescan for joint simulation. Write LQR control code in matlab, carsim provides dynamic model, prescan provides simulation scene and positioning, and the maximum vehicle speed is set to 1m / s.

[0125] Fig.24 The figure shows the comparison between the simulated path and the theoretical path for parallel parking. Fig.25 This is a comparison chart of the simulated heading angle and the theoretical heading angle for the first path in parallel parking. Fig.26 This is a comparison diagram of the simulated heading angle and the theoretical heading angle for the second path in parallel parking.

[0126] Reference Figure 24 to Figure 26 It can be seen that when the more classic LQR algorithm is used for path tracking, the parallel parking tracking error is no more than 0.065m, the heading angle error is no more than 0.037rad, and there is no collision between the vehicle and the road boundary and the parking space boundary during the entire simulation. It can be seen that the parallel parking path calculated by the improved dung beetle optimization algorithm is more reasonable.

[0127] Fig. 27 The figure shows the comparison between the simulated path and the theoretical path for perpendicular parking. Fig.28 A comparison chart of the simulated heading angle and the theoretical heading angle for the first path in perpendicular parking. Fig.29 The figure is a comparison diagram between the simulated heading angle and the theoretical heading angle of the second path in perpendicular parking.

[0128] Reference Figure 27 to Figure 29 It can be seen that when the more classic LQR algorithm is used for path tracking, the tracking error of the vertical parking process is no more than 0.071m, and the heading angle error is no more than 0.043rad. During the entire simulation process, there is no collision between the vehicle and the road boundary and the parking space boundary. It can be seen that the vertical parking path calculated by the improved dung beetle optimization algorithm is more reasonable.

[0129] In an embodiment of the present invention, a multi-constraint parking path planning device is also provided, comprising:

[0130] The position optimization module is used to determine multiple candidate parking start positions and parking end positions of the vehicle based on the dung beetle optimization algorithm and the current position of the vehicle. The path of the vehicle moving from the current position to the parking start position is the first path, and the path of the vehicle moving from the current position to the parking end position is the second path.

[0131] The path determination module is used to determine the target parking start position and target parking end position of the vehicle and the corresponding first path and second path from a plurality of candidate parking start positions and parking end positions of the vehicle through a target optimization model.

[0132] The target optimization model includes an objective function and optimization constraints. The objective function takes minimizing the total length of the first path and the second path as the optimization goal. The optimization constraints include a collision constraint between the first path and the second path and a curvature constraint of the second path.

[0133] In an embodiment of the present invention, a vehicle control system is further provided, comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the multi-constraint parking path planning method provided by the present invention.

[0134] In an embodiment of the present invention, a computer-readable storage medium is further provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the multi-constraint parking path planning method provided by the present invention are implemented.

[0135] It should be understood that the specific embodiments described herein are only used to explain the application, rather than to limit it. Based on the embodiments provided in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of this application.

[0136] Obviously, the drawings are only some examples or embodiments of the present application. For ordinary technicians in the field, the present application can also be applied to other similar situations based on these drawings without creative work. In addition, it is understandable that although the work done in this development process may be complicated and lengthy, for ordinary technicians in the field, certain changes in design, manufacturing or production based on the technical content disclosed in this application are only conventional technical means and should not be regarded as insufficient content disclosed in this application.

Claims

1. A multi-constraint parking path planning method, characterized in that: include: Based on the current position of the vehicle, determining at least one candidate parking start position and parking end position of the vehicle through an improved rolling ball dung beetle strategy; Based on the current position of the vehicle, determining at least one candidate parking start position and parking end position of the vehicle through an improved egg ball breeding strategy; Based on the current position of the ego vehicle, determining at least one candidate parking start position and parking end position of the ego vehicle through an improved foraging dung beetle strategy; The path of the vehicle moving from the current position to the parking start position is the first path, and the path of the vehicle moving from the current position to the parking end position is the second path; Determining a target parking start position and a target parking end position of the vehicle and a corresponding first path and a second path from a plurality of candidate parking start positions and parking end positions of the vehicle through a target optimization model; The target optimization model includes an objective function and optimization constraints. The objective function takes minimizing the total length of the first path and the second path as the optimization goal. The optimization constraints include a collision constraint between the first path and the second path and a curvature constraint of the second path.

2. The multi-constraint parking path planning method according to claim 1, characterized in that: Improved rolling ball dung beetle strategies include: ， in, r 1. r 2. n 1 and n 2 represent different parameters respectively. a 1 and a 2 represents a random number between 0 and 1, t represents the number of iterations, X i ( t ) indicates the t The iteration i The location of the dung beetle, X i ( t+ 1) Indicates the t +1 iteration i The location of the dung beetle, bestX It represents the optimal individual position in the current dung beetle population. Dung beetles are used to represent the vehicle.

3. The multi-constraint parking path planning method according to claim 1, characterized in that: Improved egg ball propagation strategies include: ， in, t represents the number of iterations, X i ( t ) indicates the t The first iteration i The location of the dung beetle, X i ( t+ 1) Indicates the t +1 iteration i The location of the dung beetle, bestX represents the optimal individual position in the current dung beetle population. Dung beetles are used to represent the vehicle. r Indicates the parameters, g is a constant, M represents the maximum number of iterations, l Represents a random number between -1 and 1. Lb * and Ub * They represent the lower and upper limits of the dung beetle egg-laying area respectively. The dung beetle egg-laying area represents the area where the vehicle can stay.

4. The multi-constraint parking path planning method according to claim 1, characterized in that: Improved foraging strategies for dung beetles include: ， in, t represents the number of iterations, X i ( t ) indicates the t The first iteration i The location of the dung beetle, X i ( t+ 1) Indicates the t +1 iteration i The location of the dung beetle, bestX represents the optimal individual position in the current dung beetle population. Dung beetles are used to represent the vehicle. C 1 represents a random number that follows a normal distribution. C 2 represents a random vector belonging to (0, 1), Lb b and Ub b They represent the lower and upper limits of the dung beetle foraging area, respectively. The dung beetle foraging area represents the area where the vehicle can stay. M Indicates the maximum number of iterations.

5. The multi-constraint parking path planning method according to claim 1, characterized in that: The objective function is: ， in, d 1 and d 2 represents the total number of sampling points of the first path and the second path respectively, x i Indicates the first path i The horizontal coordinate of the sampling point, x i-1 Indicates the first path i -1 The horizontal coordinate of the sampling point, F 1 represents the first path function, x j Indicates the second path j The horizontal coordinate of the sampling point, x j-1 Indicates the second path j -1 The horizontal coordinate of the sampling point, F 2 represents the second path function; The curvature constraint is used to ensure that the turning radius of the ego vehicle is greater than the minimum turning radius; The collision constraint includes a collision constraint of a first path and a collision constraint of a second path; Assuming the parking space is below the lane where the vehicle is located: The collision constraint of the first path is used to ensure that the ego vehicle does not collide with the upper and lower boundaries of the lane; In the parallel parking condition, the collision constraint of the second path is used to ensure that the ego vehicle does not collide with the upper right corner and the lower boundary of the parking space; In the perpendicular parking condition, the collision constraint of the second path is used to ensure that the ego vehicle does not collide with the upper right corner and left boundary of the parking space.

6. The multi-constraint parking path planning method according to claim 5, characterized in that: The first path function is: ， in, a 0. a 1. a 2. a 3. a 4 and a 5 represent different coefficients, x 0 and y 0 represents the horizontal and vertical coordinates of the current position of the vehicle, x a and y a They represent the horizontal and vertical coordinates of the parking starting position of the vehicle, θ Indicates the current heading angle of the vehicle; The second path function is: ， In parallel parking maneuvers: ， In a perpendicular parking maneuver: ， in, b 0. b 1. b 2. b 3. b 4 and b 5 represent different coefficients, x b and y b They respectively represent the horizontal and vertical coordinates of the parking end position of the vehicle.

7. A multi-constraint parking path planning device, characterized in that: include: a position optimization module, for determining at least one candidate parking start position and parking end position of the vehicle based on the current position of the vehicle by using an improved rolling ball dung beetle strategy, determining at least one candidate parking start position and parking end position of the vehicle by using an improved egg ball breeding strategy, and determining at least one candidate parking start position and parking end position of the vehicle by using an improved foraging dung beetle strategy, wherein the path of the vehicle moving from the current position to the parking start position is a first path, and the path of the vehicle moving from the current position to the parking end position is a second path; A path determination module, used to determine a target parking start position and a target parking end position of the vehicle and a corresponding first path and a second path from a plurality of candidate parking start positions and parking end positions of the vehicle through a target optimization model; The target optimization model includes an objective function and optimization constraints. The objective function takes minimizing the total length of the first path and the second path as the optimization goal. The optimization constraints include a collision constraint between the first path and the second path and a curvature constraint of the second path.

8. An automobile control system, comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to execute the multi-constraint parking path planning method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the multi-constraint parking path planning method according to any one of claims 1 to 6 are implemented.

Images