Improved hybrid A star path planning method for autonomous vehicle

Abstract

The invention discloses an improved hybrid A star path planning method for an automatic driving vehicle, and relates to the technical field of automatic driving, laser radar sensing and unmanned vehicle navigation. According to the method, on the basis of inheriting the advantages of high global search efficiency, strong kinematics feasibility and the like of a hybrid A star algorithm, a multi-factor cost function is designed, a trajectory smooth constraint is introduced, and the feasibility and execution quality of a path are improved from two aspects of cost evaluation and path construction. In order to further enhance the smoothness and the dynamic controllability of the local path, the initial path is optimized and adjusted by fusing a time elastic band (TEB) algorithm in the method, so that the continuity and the stability of the path in a complex dynamic environment are improved. The method is suitable for a high-precision path planning task in an automatic driving system, and the comprehensive requirements of a path generation result in the aspects of reliability and environmental adaptability can be well met.

Description

Technical Field

[0001] The present invention relates to the fields of autonomous driving, laser radar and unmanned vehicle navigation technology, and in particular to an improved hybrid A-star path planning method for autonomous driving vehicles. Background Art

[0002] With the continuous development of autonomous driving technology, the requirements for the efficiency, smoothness, and controllability of path planning in autonomous navigation systems are increasing. How to achieve efficient, smooth, and controllable path planning has become one of the key technologies in autonomous navigation systems. Path planning is not only related to the vehicle's ability to safely avoid obstacles in complex environments, but also directly affects the controllability and execution stability of its motion trajectory. Although traditional path planning methods (such as the A-star algorithm and the RRT algorithm) have good stability and adaptability in relatively simple or open environments, they often face problems such as uneven paths, sharp turns, redundant nodes, and long planning times in complex scenarios such as confined space and dense obstacles. These problems make it difficult to meet the comprehensive requirements of autonomous driving systems for path continuity, smoothness, and dynamic executability.

[0003] To alleviate the above problems, the hybrid A-star algorithm introduces a nonholonomic kinematic model of the vehicle and incorporates kinematic constraints into the search process, thereby achieving preliminary satisfaction of feasibility conditions in path planning. While inheriting the global search capabilities of the traditional A-star algorithm, this method can generate an initial path with kinematic feasibility. Therefore, it has become one of the important improvement strategies widely used in the field of autonomous driving in recent years. However, the current hybrid A-star method still faces many challenges in practical application, such as discontinuous changes in path angles, limited discrete accuracy of front wheel angles, and the lack of perception and adaptability of heuristic functions to environmental structure. As a result, the generated path still has deficiencies in local controllability, smoothness, and robustness. Therefore, existing path planning methods are insufficient in terms of path smoothness and feasibility.

[0004] In addition, most traditional path cost evaluation models are based on Euclidean distance or simple heuristic functions, which fail to fully consider multi-dimensional factors such as the local environment geometry, obstacle distribution density, path curvature changes, and vehicle dynamic constraints. It is difficult to accurately evaluate path quality, resulting in a lack of flexibility and adaptability when facing dynamic obstacles, asymmetric boundaries, or sharp turns. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to overcome the shortcomings of the existing technology and provide an improved hybrid A-star path planning method for autonomous vehicles. This method introduces a high-dimensional state vector consisting of vehicle position, front wheel steering angle and heading angle, and combines the vehicle's non-holonomic kinematic constraints to expand the path; at the same time, a multi-index weighted cost function is designed to evaluate candidate paths to determine the optimal path, and then the time elastic band (TEB) algorithm is combined to optimize and adjust the optimal path, thereby significantly improving the smoothness and executability of path planning.

[0006] The present invention adopts the following technical solutions to solve the above technical problems:

[0007] An improved hybrid A-star path planning method for an autonomous driving vehicle proposed in the present invention includes:

[0008] Build environmental raster maps;

[0009] A high-dimensional state vector including the vehicle position, front wheel steering angle, and heading angle is introduced to fully describe the vehicle's motion state in the environment grid map.

[0010] In the path search phase, based on the high-dimensional state vector, combined with the vehicle's non-holonomic constraints and relying on the spatial information of the environment grid map, multiple candidate paths that meet kinematic feasibility are generated;

[0011] A heuristic function is constructed that combines the cost of the shortest single-source path with the Reeds-Shepp distance of the shortest curved path that allows bidirectional travel. This heuristic function is used to guide the path to converge to the target area within an environmental grid map. Candidate paths are comprehensively evaluated within the environmental grid map to select the optimal path.

[0012] Combined with the spatial layout in the environmental grid map, the optimal path is smoothly optimized to generate a navigation trajectory.

[0013] As a further optimization scheme for the improved hybrid A-star path planning method for autonomous vehicles described in the present invention, the specific steps are as follows:

[0014] Step 1: Obtain the spatial information of the target environment and use the grid mapping algorithm to construct an environmental grid map. The environmental grid map is used to represent the distribution of obstacles and the traversable area.

[0015] Step 2: In the state-space modeling process, a high-dimensional state vector including vehicle position, front wheel steering angle, and heading angle is introduced to fully describe the vehicle's motion state, thereby ensuring the kinematic feasibility of the path generation process.

[0016] Step 3: Based on the high-dimensional state vector constructed in step 2, multiple feasible candidate paths are generated in the path extension process in combination with the vehicle's non-holonomic constraints;

[0017] Step 4: Construct a heuristic function that combines the cost of the shortest single-source path and the Reeds-Shepp distance of the shortest curved path that allows bidirectional travel. Use the heuristic function to guide the path to converge to the target area in the environmental grid map.

[0018] Step 5: In the path evaluation phase, a weighted cost function is constructed to evaluate candidate paths based on path length, curvature change, vehicle direction switching frequency, and obstacle boundary distance to obtain the optimal path.

[0019] Step 6: Smooth the optimal path to generate the final navigation trajectory.

[0020] As a further optimization scheme of the improved hybrid A-star path planning method for autonomous driving vehicles described in the present invention, the single-source shortest path cost is calculated by the Dijkstra algorithm, and the optimal path is smoothed by the time elastic band algorithm.

[0021] As a further optimization scheme of the improved hybrid A-star path planning method for autonomous vehicles described in the present invention, spatial information of the target environment is acquired through a lidar or camera, and an environment grid map is constructed using a grid mapping algorithm. The environment grid map is used to represent the distribution of obstacles and traversable areas. The details are as follows:

[0022] Obtain environmental spatial data through lidar or camera, divide the environmental spatial data into fixed-size grid cells, and recursively update the occupancy probability of each grid based on the Bayesian formula to represent the obstacle distribution and passage area;

[0023] Assume that a grid cell in the environmental spatial data is m, given the sensor observation z t and sensor pose x t , the posterior probability that the grid cell is occupied by an obstacle is expressed as:

[0024]

[0025] Among them, P(m i |z 1:t ,x 1:t ) is the posterior probability of the grid cell being occupied at the current moment; P(z t |m i ,x t ) is the observed z at the current position and when the grid state is occupied t The probability of P(m i |z 1:t-1 ,x1:t-1 ) is the prior probability of the previous moment; the normalization factor P(z t |z 1:t-1 ,x 1:t ) is calculated using the total probability formula.

[0026] As a further optimization scheme of the improved hybrid A-star path planning method for autonomous vehicles according to the present invention, the high-dimensional state vector s in step 2 is expressed as:

[0027] s=[x,y,θ,d]

[0028] Where x and y represent the position coordinates of the vehicle in two-dimensional space; θ represents the vehicle's heading angle; and the variable d represents the vehicle's movement: when d = +1, it indicates the vehicle is moving forward; when d = -1, it indicates the vehicle is moving backward.

[0029] As a further optimization scheme of the improved hybrid A-star path planning method for autonomous vehicles described in the present invention, based on the high-dimensional state vector constructed in step 2, multiple feasible candidate paths are generated in combination with the vehicle's nonholonomic constraints during the path extension process; wherein the extension method is:

[0030] x i+1 =x i +d i ·Δs·cosθ i+1

[0031] y i+1 =y i +d i ·Δs·cosθ i+1

[0032] θ i+1 =M(θ i +d i ·Δs·tan(Δδ i ) / L)

[0033] Among them, x i with y i Represents the plane coordinate position of the vehicle in the i-th state, θ i represents the orientation angle of the vehicle in the i-th state, that is, the angle between the vehicle's forward direction and the X-axis of the coordinate system; x i+1 、y i+1 ,θ i+1 Respectively represent the plane coordinate position and orientation angle of the vehicle in the next state i+1; variable d i A value of +1 indicates that the vehicle is moving forward, and a value of -1 indicates that the vehicle is moving backward; Δs represents the step length of the vehicle when the current path is extended; Δδ irepresents the change in the front wheel angle under the step size of Δs; L represents the wheelbase of the vehicle, that is, the distance between the front wheel axle and the rear wheel axle, and the function M(·) represents the angle normalization operation, which is used to convert the heading angle θ i and θ i+1 Keep in the range [-π,π].

[0034] As a further optimization scheme of the improved hybrid A-star path planning method for autonomous vehicles described in the present invention, the heuristic function h(n) in step 4 is:

[0035] h(n)=α·h D (n)+β·h RS (n)

[0036] Among them, h D (n) represents the Dijkstra shortest distance estimate between the current node and the target point, which is used to reflect the optimal path cost when motion constraints are not considered on the environment grid map; h RS (n) represents the length of the trajectory that satisfies the kinematic constraints and is generated based on the current state using the Reeds-Shepp curve. It is used to compensate for the insufficient consideration of vehicle accessibility by the heuristic function. α and β are weight coefficients, and the values of α and β parameters are adjusted according to the specific scenario.

[0037] As a further optimization scheme of the improved hybrid A-star path planning method for autonomous driving vehicles described in the present invention, the weighted cost function is:

[0038] Let path I be a series of nodes n1, n2, ..n k , where n k' For the k'th node, 1≤k'≤k, the weighted cost function C(I) of the path is defined as the weighted sum of multiple evaluation factors:

[0039] C(I)=w1·C length (I)+w2·C curvature (I)+w3·C turning (I)+w4·C obstacle (I)

[0040] Among them, C length (I) is the path length cost, w1 is the path length trade-off factor, C curvature (I) is the curvature change cost, w2 is the curvature change trade-off factor, C turning (I) is the vehicle direction switching frequency cost, w3 is the vehicle direction switching frequency trade-off factor, C obstacle (I) is the obstacle boundary distance cost, and w4 is the obstacle boundary distance trade-off factor.

[0041] As a further optimization scheme of the improved hybrid A-star path planning method for autonomous vehicles described in the present invention, a time elastic band algorithm is used to smooth the optimal path. The time elastic band algorithm minimizes the following cost function:

[0042]

[0043] Among them, J is the constructed cost function, and N is the number of discrete nodes in the path; is the path smoothing cost, w s is the weight coefficient of the path smoothing term; is the kinematic cost term, w d is the kinematics weight coefficient, is the obstacle avoidance cost, w o is the weight coefficient of obstacle avoidance item; is the expected speed cost, w v is the weight coefficient of the expected speed term; is the acceleration constraint cost, w a is the weight coefficient of the acceleration constraint term.

[0044] A computer device includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the computer program, the steps of the improved hybrid A-star path planning method for autonomous vehicles as described above are implemented.

[0045] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0046] (1) A cost function that comprehensively considers multiple path performance indicators is designed to effectively balance path length, turning changes, and obstacle spacing factors during the path planning process. This significantly improves the overall efficiency and practicality of path generation, especially in environments with high path complexity or dense obstacles, and improves the path redundancy and lack of security problems existing in the traditional hybrid A-star method.

[0047] (2) A heuristic function is constructed that integrates the single-source shortest path cost (Dijkstra) and the shortest curve path distance for bidirectional travel (Reeds-Shepp). During the search process, the front wheel angle limit, the minimum turning radius constraint, and the path smoothness requirement are comprehensively considered. This function is particularly suitable for scenarios with high requirements for path continuity and kinematic feasibility, and effectively guides the search to converge to the target area, thereby improving the computational efficiency of path planning.

[0048] (3) A local trajectory optimization strategy based on the time elastic band (TEB) algorithm is introduced to further improve the smoothness and dynamic adjustment capability of the path in the local area in the presence of dynamic obstacles or local space constraints, thereby enhancing the controllability and stability of the path during actual execution. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 Flowchart of the present invention.

[0050] Figure 2 Flowchart for vehicle path search. DETAILED DESCRIPTION

[0051] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0052] Based on the hybrid A-star algorithm, this paper introduces a more refined cost evaluation mechanism, supplemented by an appropriate local trajectory optimization strategy, to improve the overall performance of the path in terms of smoothness, feasibility, and execution stability, and better meet the requirements of autonomous driving systems for high-reliability and high-quality generation of navigation paths in highly constrained and complex environments.

[0053] Figure 1 This is a flowchart of the present invention. The core idea of the present invention is to propose an improved hybrid A-star path planning method for autonomous vehicles. System programming is carried out according to the theoretical structure from simple to complex and from bottom to top. The specific implementation method is as follows:

[0054] 1. First, the environment space information is acquired through lidar or camera, and the space is divided into grid cells of fixed size; then, the occupancy probability of each grid cell is recursively updated based on the Bayesian formula to reflect the distribution of obstacles and the passable area, thereby constructing the environment grid map. Specifically, let a grid cell in the environment space data be m, given the sensor observation z t and sensor pose x t , the posterior probability that the grid cell is occupied by an obstacle is expressed as:

[0055]

[0056] Among them, P(m i |z 1:t ,x 1:t ) is the posterior probability of the grid cell being occupied at the current moment; P(z t |m i ,x t ) is the observed z at the current position and when the grid state is occupied t The probability of P(m i |z1:t-1 ,x 1:t-1 ) is the prior probability of the previous moment; the normalization factor P(z t |z 1:t-1 ,x 1:t ) is calculated using the total probability formula.

[0057] 2. In the state-space modeling process, a high-dimensional state vector consisting of vehicle position, front wheel steering angle, and heading angle is introduced to comprehensively describe the vehicle's motion state, thereby ensuring the kinematic feasibility of the path generation process.

[0058] 3. In the path search process, such as Figure 2 The figure shows the vehicle path search flow chart. Based on the constructed high-dimensional state vector and combined with the vehicle's nonholonomic constraints, multiple candidate paths for the next step are obtained based on the current node expansion. The specific expansion method is as follows:

[0059] x i+1 =x i +d i ·Δs·cosθ i+1

[0060] y i+1 =y i +d i ·Δs·cosθ i+1

[0061] θ i+1 =M(θ i +d i ·Δs·tan(Δδ i ) / L)

[0062] Among them, x i with y i Represents the plane coordinate position of the vehicle in the i-th state, θ i Indicates the vehicle's heading angle in this state, that is, the angle between the vehicle's forward direction and the X-axis of the coordinate system; x i+1 、y i+1 ,θ i+1 Respectively represent the plane coordinate position and orientation angle of the vehicle in the next state i+1. i A value of +1 indicates that the vehicle is moving forward, and a value of -1 indicates that the vehicle is moving backward; Δs represents the step length of the vehicle when the current path is extended; Δδ i represents the change in the front wheel angle under the step size of Δs; L represents the wheelbase of the vehicle, that is, the distance between the front wheel axle and the rear wheel axle. The function M(·) represents the angle normalization operation, which is used to convert the heading angle θ i and θ i+1 Keep in the range [-π,π].

[0063] 4. To improve the efficiency of path search, a heuristic function is constructed that combines the cost of the single-source shortest path search algorithm (Dijkstra) and the distance of the shortest curve path (Reeds-Shepp) that allows bidirectional travel. The heuristic function h(n) is expressed as:

[0064] h(n)=α·h D (n)+β·h RS (n)

[0065] Among them, h D (n) represents the Dijkstra shortest distance estimate between the current node and the target point, which is used to reflect the optimal path cost when motion constraints are not considered on the environment grid map; h RS (n) represents the length of the trajectory that satisfies the kinematic constraints and is generated based on the current state using the Reeds-Shepp curve. It is used to compensate for the insufficient consideration of vehicle accessibility by the heuristic function. α and β are weight coefficients, and the values of α and β parameters are adjusted according to the specific scenario.

[0066] 5. A weighted cost function is constructed to evaluate candidate paths by comprehensively considering path length, curvature change, vehicle direction switching frequency, and obstacle boundary distance to obtain the optimal path. Specifically, path I is set as a series of nodes n1, n2, ..n k , where n k' For the k'th node, 1≤k'≤k, the weighted cost function C(I) of the path is defined as the weighted sum of multiple evaluation factors:

[0067] C(I)=w1·C length (I)+w2·C curvature (I)+w3·C turning (I)+w4·C obstacle (I)

[0068] Among them, C length (I) is the path length cost, w1 is the path length trade-off factor, C curvature (I) is the curvature change cost, w2 is the curvature change trade-off factor, C turning (I) is the vehicle direction switching frequency cost, w3 is the vehicle direction switching frequency trade-off factor, C obstacle (I) is the obstacle boundary distance cost, and w4 is the obstacle boundary distance trade-off factor. Adjustments are made based on the weight coefficients w1, w2, w3, and w4 to balance the length, smoothness, directional stability, and safety of the path, ultimately selecting a path with the best overall performance.

[0069] 6. During the path optimization phase, a time elastic band algorithm is introduced to smoothly optimize the optimal path. By constructing an overall objective optimization function that incorporates factors such as path smoothness, dynamic feasibility, obstacle avoidance, and speed rationality, and combining it with nonlinear optimization methods, the positions and time intervals of local paths are dynamically adjusted.

[0070] An embodiment of the present invention also provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the computer program, the steps of the improved hybrid A-star path planning method for autonomous driving vehicles as described above are implemented.

[0071] An embodiment of the present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the improved hybrid A-star path planning method for autonomous driving vehicles as described above.

[0072] It will be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention may be implemented in various computer languages, for example, the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0073] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0074] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0075] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0076] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0077] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. An improved hybrid A-star path planning method for autonomous driving vehicles, characterized in that: include: Build environmental raster maps; A high-dimensional state vector including vehicle position, front wheel steering angle, and heading angle is introduced to fully describe the vehicle's motion state in the environment grid map. In the path search phase, based on the high-dimensional state vector, combined with the vehicle's non-holonomic constraints and relying on the spatial information of the environment grid map, multiple candidate paths that meet kinematic feasibility are generated; A heuristic function is constructed that combines the cost of the shortest single-source path with the Reeds-Shepp distance of the shortest curved path that allows bidirectional travel. This heuristic function is used to guide the path to converge to the target area within an environmental grid map. Candidate paths are comprehensively evaluated within the environmental grid map to select the optimal path. Combined with the spatial layout in the environmental grid map, the optimal path is smoothly optimized to generate a navigation trajectory.

2. The improved hybrid A-star path planning method for autonomous driving vehicles according to claim 1, characterized in that: The specific steps are as follows: Step 1: Obtain the spatial information of the target environment and use the grid mapping algorithm to construct an environmental grid map. The environmental grid map is used to represent the distribution of obstacles and the traversable area. Step 2: In the state-space modeling process, a high-dimensional state vector including vehicle position, front wheel steering angle, and heading angle is introduced to fully describe the vehicle's motion state, thereby ensuring the kinematic feasibility of the path generation process. Step 3: Based on the high-dimensional state vector constructed in step 2, multiple feasible candidate paths are generated in the path extension process in combination with the vehicle's non-holonomic constraints; Step 4: Construct a heuristic function that combines the cost of the shortest single-source path and the Reeds-Shepp distance of the shortest curved path that allows bidirectional travel. Use the heuristic function to guide the path to converge to the target area in the environmental grid map. Step 5: In the path evaluation phase, a weighted cost function is constructed to evaluate candidate paths based on path length, curvature change, vehicle direction switching frequency, and obstacle boundary distance to obtain the optimal path. Step 6: Smooth the optimal path to generate the final navigation trajectory.

3. The improved hybrid A-star path planning method for autonomous driving vehicles according to claim 1, characterized in that: The single-source shortest path cost is calculated using the Dijkstra algorithm, and the optimal path is smoothed using the time elastic band algorithm.

4. The improved hybrid A-star path planning method for autonomous driving vehicles according to claim 2, characterized in that: The spatial information of the target environment is obtained through lidar or camera, and the grid mapping algorithm is used to construct an environmental grid map. The environmental grid map is used to represent the distribution of obstacles and the traversable area. The details are as follows: Obtain environmental spatial data through lidar or camera, divide the environmental spatial data into fixed-size grid cells, and recursively update the occupancy probability of each grid based on the Bayesian formula to represent the obstacle distribution and passage area; Assume that a grid cell in the environmental spatial data is m, given the sensor observation z t and sensor pose x t , the posterior probability that the grid cell is occupied by an obstacle is expressed as: Among them, P(m i |z 1:t ,x 1:t ) is the posterior probability of the grid cell being occupied at the current moment; P(z t |m i ,x t ) is the observed z at the current position and when the grid state is occupied t The probability of P(m i |z 1:t-1 ,x 1:t-1 ) is the prior probability of the previous moment; the normalization factor P(z t |z 1:t-1 ,x 1:t ) is calculated using the total probability formula.

5. The improved hybrid A-star path planning method for autonomous driving vehicles according to claim 2, characterized in that: The high-dimensional state vector s in step 2 is expressed as: s=[x,y,θ,d] Where x and y represent the position coordinates of the vehicle in two-dimensional space; θ represents the vehicle's heading angle; and the variable d represents the vehicle's movement: when d = +1, it indicates the vehicle is moving forward; when d = -1, it indicates the vehicle is moving backward.

6. The improved hybrid A-star path planning method for autonomous driving vehicles according to claim 2, characterized in that: Based on the high-dimensional state vector constructed in step 2, multiple feasible candidate paths are generated in the path extension process by combining the vehicle's nonholonomic constraints. The extension method is: x i+1 =x i +d i ·Δs·cosθ i+1 y i+1 =y i +d i ·Δs·cosθ i+1 i i+1 =M(θ i +d i ·Δs·tan(Δδ i ) / L) Among them, x i with y i Represents the plane coordinate position of the vehicle in the i-th state, θ i represents the orientation angle of the vehicle in the i-th state, that is, the angle between the vehicle's forward direction and the X-axis of the coordinate system; x i+1 、y i+1 ,θ i+1 Respectively represent the plane coordinate position and orientation angle of the vehicle in the next state i+1; variable d i A value of +1 indicates that the vehicle is moving forward, and a value of -1 indicates that the vehicle is moving backward; Δs represents the step length of the vehicle when the current path is extended; Δδ i represents the change in the front wheel angle under the step size of Δs; L represents the wheelbase of the vehicle, that is, the distance between the front wheel axle and the rear wheel axle, and the function M(·) represents the angle normalization operation, which is used to convert the heading angle θ i and θ i+1 Keep in the range [-π,π].

7. The improved hybrid A-star path planning method for autonomous driving vehicles according to claim 2, characterized in that: The heuristic function h(n) in step 4 is: h(n)=α·h D (n)+β·h RS (n) Among them, h D (n) represents the Dijkstra shortest distance estimate between the current node and the target point, which is used to reflect the optimal path cost when motion constraints are not considered on the environment grid map; h RS (n) represents the length of the trajectory that satisfies the kinematic constraints and is generated based on the current state using the Reeds-Shepp curve. It is used to compensate for the insufficient consideration of vehicle accessibility by the heuristic function. α and β are weight coefficients, and the values of α and β parameters are adjusted according to the specific scenario.

8. The improved hybrid A-star path planning method for autonomous driving vehicles according to claim 2, characterized in that: The weighted cost function is: Let path I be a series of nodes n1, n2, ..n k , where n k ' is the k'th node, 1≤k'≤k, and the weighted cost function C(I) of the path is defined as the weighted sum of multiple evaluation factors: C(I)=w1·C length (I)+w2·C curvature (I)+w3·C turning (I)+w4·C obstacle (I) Among them, C length (I) is the path length cost, w1 is the path length trade-off factor, C curvature (I) is the curvature change cost, w2 is the curvature change trade-off factor, C turning (I) is the vehicle direction switching frequency cost, w3 is the vehicle direction switching frequency trade-off factor, C obstacle (I) is the obstacle boundary distance cost, and w4 is the obstacle boundary distance trade-off factor.

9. The improved hybrid A-star path planning method for autonomous driving vehicles according to claim 2, characterized in that: The time elastic band algorithm is used to smooth the optimal path. The time elastic band algorithm minimizes the following cost function: Among them, J is the constructed cost function, and N is the number of discrete nodes in the path; is the path smoothing cost, w s is the weight coefficient of the path smoothing term; is the kinematic cost term, w d is the kinematics weight coefficient, is the obstacle avoidance cost, w o is the weight coefficient of obstacle avoidance item; is the expected speed cost, w v is the weight coefficient of the expected speed term; is the acceleration constraint cost, w a is the weight coefficient of the acceleration constraint term.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that: When the processor executes the computer program, the steps of the improved hybrid A-star path planning method for autonomous driving vehicles as described in any one of claims 1 to 9 are implemented.

Images