A method and system for obstacle avoidance by unmanned vehicles

Abstract

This invention discloses an obstacle avoidance method and system for autonomous vehicles. Based on the tracking path, the current tracking target position is determined according to the autonomous vehicle's driving information. If an obstacle exists, a priori static map is segmented based on the driving information and the current tracking target position to obtain a sub-map. The obstacle is then mapped into the sub-map to obtain an obstacle distance information map. Based on the obstacle distance information map, a hybrid A* path planning method is used to plan the path and obtain the obstacle avoidance path. Thus, when an obstacle is detected, subsequent path planning is performed within the sub-map range, which greatly improves the efficiency of path planning and can plan a suitable path for the vehicle more quickly. It can flexibly plan the obstacle avoidance path according to the specific location and distance information of the obstacle, enabling the vehicle to better adapt to complex and changing driving environments, thereby improving the real-time performance of obstacle avoidance for autonomous vehicles.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and in particular to a method and system for autonomous vehicles to avoid obstacles. Background Technology

[0002] Currently, autonomous vehicles can travel along pre-defined paths. These paths are typically pre-collected or planned via mouse clicks on the vehicle monitoring map interface of the control platform before being sent to the autonomous vehicle. During actual path-following, obstacles may appear on the path due to environmental changes, requiring obstacle avoidance algorithms to ensure the vehicle navigates around them and continues its journey. However, existing path planning algorithms are primarily global path planning adapted to static maps, which cannot guarantee real-time performance. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a method and system for unmanned vehicles to avoid obstacles, which can improve the real-time performance of unmanned vehicles in avoiding obstacles.

[0004] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0005] A method for obstacle avoidance by an autonomous vehicle includes the following steps:

[0006] Receive a prior static map of the autonomous vehicle and the tracking path on the prior static map;

[0007] The current target location is determined based on the tracking path and the driving information of the autonomous vehicle.

[0008] Determine whether there is an obstacle in front of the autonomous vehicle. If so, segment the prior static map based on the driving information and the current tracking target position to obtain a sub-map, and map the obstacle into the sub-map to obtain an obstacle distance information map.

[0009] Based on the obstacle distance information map, the hybrid A-star path planning method is used to plan the path of the unmanned vehicle to obtain the obstacle avoidance path.

[0010] To solve the above-mentioned technical problems, another technical solution adopted by the present invention is as follows:

[0011] An obstacle avoidance system for unmanned vehicles includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it performs the following steps:

[0012] Receive a prior static map of the autonomous vehicle and the tracking path on the prior static map;

[0013] The current target location is determined based on the tracking path and the driving information of the autonomous vehicle.

[0014] Determine whether there is an obstacle in front of the autonomous vehicle. If so, segment the prior static map based on the driving information and the current tracking target position to obtain a sub-map, and map the obstacle into the sub-map to obtain an obstacle distance information map.

[0015] Based on the obstacle distance information map, the hybrid A-star path planning method is used to plan the path of the unmanned vehicle to obtain the obstacle avoidance path.

[0016] The beneficial effects of this invention are as follows: Based on the tracking path, the current tracking target position is determined according to the driving information of the autonomous vehicle. If there is an obstacle in front of the autonomous vehicle, the prior static map is segmented based on the driving information and the current tracking target position to obtain a sub-map. The obstacle is then mapped into the sub-map to obtain an obstacle distance information map. Based on the obstacle distance information map, a hybrid A-star path planning method is used to plan the path for the autonomous vehicle to obtain an obstacle avoidance path. Thus, when the autonomous vehicle detects an obstacle, the prior static map is segmented to obtain a sub-map, and subsequent path planning is performed within the sub-map area, which greatly improves the efficiency of path planning and can plan a suitable path for the vehicle more quickly. The path planning based on the obstacle distance information map using the hybrid A-star path planning method can flexibly plan obstacle avoidance paths according to the specific location and distance information of the obstacle, enabling the vehicle to better adapt to complex and changing driving environments and react promptly to suddenly appearing obstacles, thereby improving the real-time performance of obstacle avoidance for the autonomous vehicle. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the steps of an obstacle avoidance method for an unmanned vehicle according to an embodiment of the present invention.

[0018] Figure 2 This is a schematic diagram of the structure of an unmanned vehicle obstacle avoidance system according to an embodiment of the present invention;

[0019] Figure 3 This is a schematic diagram of sub-map segmentation in an obstacle avoidance method for an unmanned vehicle according to an embodiment of the present invention;

[0020] Figure 4 This is a flowchart illustrating the obstacle avoidance planning process in an obstacle avoidance method for an unmanned vehicle according to an embodiment of the present invention. Detailed Implementation

[0021] To explain in detail the technical content, objectives, and effects of the present invention, the following description is provided in conjunction with the embodiments and accompanying drawings.

[0022] Please refer to Figure 1 A method for unmanned vehicles to avoid obstacles, comprising the following steps:

[0023] Receive a prior static map of the autonomous vehicle and the tracking path on the prior static map;

[0024] The current target location is determined based on the tracking path and the driving information of the autonomous vehicle.

[0025] Determine whether there is an obstacle in front of the autonomous vehicle. If so, segment the prior static map based on the driving information and the current tracking target position to obtain a sub-map, and map the obstacle into the sub-map to obtain an obstacle distance information map.

[0026] Based on the obstacle distance information map, the hybrid A-star path planning method is used to plan the path of the unmanned vehicle to obtain the obstacle avoidance path.

[0027] As can be seen from the above description, the beneficial effects of the present invention are as follows: Based on the tracking path, the current tracking target position is determined according to the driving information of the autonomous vehicle. If there is an obstacle in front of the autonomous vehicle, the prior static map is segmented based on the driving information and the current tracking target position to obtain a sub-map, and the obstacle is mapped into the sub-map to obtain an obstacle distance information map. Based on the obstacle distance information map, a hybrid A-star path planning method is used to plan the path of the autonomous vehicle to obtain an obstacle avoidance path. Thus, when the autonomous vehicle detects an obstacle, the prior static map is segmented to obtain a sub-map, and subsequent path planning is performed within the sub-map range, which greatly improves the efficiency of path planning and can plan a suitable path for the vehicle more quickly. The path planning based on the obstacle distance information map using the hybrid A-star path planning method can flexibly plan obstacle avoidance paths according to the specific location and distance information of the obstacle, enabling the vehicle to better adapt to complex and changing driving environments and react in time to suddenly appearing obstacles, thereby improving the real-time performance of obstacle avoidance of the autonomous vehicle.

[0028] Furthermore, the driving information includes the vehicle's position and driving direction;

[0029] Determining the current target location based on the tracking path and the driving information of the autonomous vehicle includes:

[0030] Obtain the pre-aiming distance of the unmanned vehicle;

[0031] Based on the vehicle position and driving direction of the autonomous vehicle, the current tracking target position, which is at a pre-aiming distance from the autonomous vehicle, is determined in the tracking path.

[0032] As described above, the tracking path itself consists of a series of waypoints. These waypoints may have overlapping points or be too densely concentrated. Based on the vehicle's position and driving direction, the current tracking target position is determined within the tracking path at a pre-aiming distance from the autonomous vehicle. This allows the vehicle sufficient time to perceive the road conditions ahead and make corresponding decisions. For example, when an obstacle is detected within the pre-aiming distance, the vehicle can plan an obstacle avoidance path or slow down to avoid collisions due to delayed reaction, thus improving the vehicle's driving safety in complex road conditions.

[0033] Furthermore, the segmentation of the prior static map based on the driving information and the current tracking target position to obtain sub-maps includes:

[0034] The road segment to be traveled is determined based on the vehicle's location and the current tracking target's location;

[0035] Based on the road segment to be traveled, determine the drivable area;

[0036] The drivable area is segmented from the prior static map to obtain a sub-map.

[0037] As described above, segmenting the drivable area from the prior static map to obtain a sub-map and processing only the map portion relevant to the current driving task greatly reduces the amount of data to be processed, improves computational efficiency, and means that the system can complete map analysis and path planning operations faster, thereby improving the vehicle's response speed and enabling the vehicle to react more promptly to road conditions and potential obstacles ahead, thus enhancing the vehicle's adaptability and safety in dynamic environments.

[0038] Further, mapping the obstacles onto the sub-map to obtain an obstacle distance information map includes:

[0039] The sub-map is discretized to obtain a two-dimensional grid map, where each grid cell represents a node;

[0040] The obstacles are mapped onto the grid map to obtain an obstacle distance information map;

[0041] The obstacle avoidance path is planned by the autonomous vehicle using a hybrid A* path planning method based on the obstacle distance information map, and includes the following:

[0042] An initial node search set is obtained from the obstacle distance information map, the node search set including nodes of the vehicle's location;

[0043] Calculate the cost value of each node in the node search set, and determine the node with the minimum cost value from the node search set based on the cost value;

[0044] Add the node with the minimum cost value to the trajectory point set;

[0045] Based on vehicle kinematics, multiple extended nodes are generated according to the node with the minimum cost value;

[0046] For each of the extended nodes, determine whether the extended node is located in the area of ​​the obstacle or has been added to the trajectory point set. If yes, the extended node is not processed. If no, the cost of the extended node is calculated.

[0047] If the extended node has been added to the node search set, then determine whether the calculated cost value of the extended node is less than the cost value of the extended node in the node search set. If it is less, then update the cost value of the extended node in the node search set.

[0048] If the extended node is not added to the node search set, then add the extended node to the node search set;

[0049] Determine whether the node with the current minimum cost value is the node at the current tracking target position. If yes, generate an obstacle avoidance path based on the trajectory point set. If no, return to the step of determining the node with the minimum cost value from the node search set based on the cost value.

[0050] As described above, each iteration selects the node with the lowest value in the current generation for expansion, which helps to quickly find a better path from the current vehicle position to the target position, reduces the time required for path planning, meets the real-time requirements of autonomous vehicles, and the final obstacle avoidance path is obtained by considering multiple factors such as obstacles, vehicle kinematics, and path value, and has high quality. It can enable the vehicle to bypass obstacles in the best way possible while ensuring safety and continuing to travel along the tracking path.

[0051] Furthermore, calculating the cost value of each node in the node search set includes:

[0052] ;

[0053] ;

[0054] ;

[0055] In the formula, f(n) represents the cost of node n, g(n) represents the cost from the node at the vehicle location to node n, h(n) represents the expected value from node n to the node at the current tracking target location, g(n-1) represents the cost from the node at the vehicle location to the previous node n-1, and L n-1,n Let represent the distance traveled from the previous node (n-1) to node n, and addcosts represent the penalty factor cost. This represents the first weighting coefficient. This represents the second weighting coefficient. L represents the third weighting coefficient. E L represents the Euclidean distance from node n to the node at the current tracking target position, without considering environmental obstacles. D L represents the Durbins curve path length from node n to the node at the current tracking target position, without considering environmental obstacles. O This represents the reciprocal of the nearest distance between node n and the obstacle.

[0056] As described above, the cost of each node consists of two parts: the historical cost (g(n)) and the expected cost (h(n)). To achieve a smoother path, a penalty factor, addcosts, is added to the historical cost calculation. The expected cost is calculated by a heuristic cost function, which consists of three parts. The weights of these three parts can be adjusted: increasing the weight of the third weight coefficient makes the search path as far away from obstacles as possible; increasing the weight of the first weight coefficient makes the search path biased towards the shortest path; and increasing the weight of the second weight coefficient makes the search path more consistent with the vehicle's kinematic trajectory, thereby ensuring the reliability of the obstacle avoidance path.

[0057] Please refer to Figure 2 An unmanned vehicle obstacle avoidance system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it performs the following steps:

[0058] Receive a prior static map of the autonomous vehicle and the tracking path on the prior static map;

[0059] The current target location is determined based on the tracking path and the driving information of the autonomous vehicle.

[0060] Determine whether there is an obstacle in front of the autonomous vehicle. If so, segment the prior static map based on the driving information and the current tracking target position to obtain a sub-map, and map the obstacle into the sub-map to obtain an obstacle distance information map.

[0061] Based on the obstacle distance information map, the hybrid A-star path planning method is used to plan the path of the unmanned vehicle to obtain the obstacle avoidance path.

[0062] As can be seen from the above description, the beneficial effects of the present invention are as follows: Based on the tracking path, the current tracking target position is determined according to the driving information of the autonomous vehicle. If there is an obstacle in front of the autonomous vehicle, the prior static map is segmented based on the driving information and the current tracking target position to obtain a sub-map, and the obstacle is mapped into the sub-map to obtain an obstacle distance information map. Based on the obstacle distance information map, a hybrid A-star path planning method is used to plan the path of the autonomous vehicle to obtain an obstacle avoidance path. Thus, when the autonomous vehicle detects an obstacle, the prior static map is segmented to obtain a sub-map, and subsequent path planning is performed within the sub-map range, which greatly improves the efficiency of path planning and can plan a suitable path for the vehicle more quickly. The path planning based on the obstacle distance information map using the hybrid A-star path planning method can flexibly plan obstacle avoidance paths according to the specific location and distance information of the obstacle, enabling the vehicle to better adapt to complex and changing driving environments and react in time to suddenly appearing obstacles, thereby improving the real-time performance of obstacle avoidance of the autonomous vehicle.

[0063] Furthermore, the driving information includes the vehicle's position and driving direction;

[0064] Determining the current target location based on the tracking path and the driving information of the autonomous vehicle includes:

[0065] Obtain the pre-aiming distance of the unmanned vehicle;

[0066] Based on the vehicle position and driving direction of the autonomous vehicle, the current tracking target position, which is at a pre-aiming distance from the autonomous vehicle, is determined in the tracking path.

[0067] As described above, the tracking path itself consists of a series of waypoints. These waypoints may have overlapping points or be too densely concentrated. Based on the vehicle's position and driving direction, the current tracking target position is determined within the tracking path at a pre-aiming distance from the autonomous vehicle. This allows the vehicle sufficient time to perceive the road conditions ahead and make corresponding decisions. For example, when an obstacle is detected within the pre-aiming distance, the vehicle can plan an obstacle avoidance path or slow down to avoid collisions due to delayed reaction, thus improving the vehicle's driving safety in complex road conditions.

[0068] Furthermore, the segmentation of the prior static map based on the driving information and the current tracking target position to obtain sub-maps includes:

[0069] The road segment to be traveled is determined based on the vehicle's location and the current tracking target's location;

[0070] Based on the road segment to be traveled, determine the drivable area;

[0071] The drivable area is segmented from the prior static map to obtain a sub-map.

[0072] As described above, segmenting the drivable area from the prior static map to obtain a sub-map and processing only the map portion relevant to the current driving task greatly reduces the amount of data to be processed, improves computational efficiency, and means that the system can complete map analysis and path planning operations faster, thereby improving the vehicle's response speed and enabling the vehicle to react more promptly to road conditions and potential obstacles ahead, thus enhancing the vehicle's adaptability and safety in dynamic environments.

[0073] Further, mapping the obstacles onto the sub-map to obtain an obstacle distance information map includes:

[0074] The sub-map is discretized to obtain a two-dimensional grid map, where each grid cell represents a node;

[0075] The obstacles are mapped onto the grid map to obtain an obstacle distance information map;

[0076] The obstacle avoidance path is planned by the autonomous vehicle using a hybrid A* path planning method based on the obstacle distance information map, and includes the following:

[0077] An initial node search set is obtained from the obstacle distance information map, the node search set including nodes of the vehicle's location;

[0078] Calculate the cost value of each node in the node search set, and determine the node with the minimum cost value from the node search set based on the cost value;

[0079] Add the node with the minimum cost value to the trajectory point set;

[0080] Based on vehicle kinematics, multiple extended nodes are generated according to the node with the minimum cost value;

[0081] For each of the extended nodes, determine whether the extended node is located in the area of ​​the obstacle or has been added to the trajectory point set. If yes, the extended node is not processed. If no, the cost of the extended node is calculated.

[0082] If the extended node has been added to the node search set, then determine whether the calculated cost value of the extended node is less than the cost value of the extended node in the node search set. If it is less, then update the cost value of the extended node in the node search set.

[0083] If the extended node is not added to the node search set, then add the extended node to the node search set;

[0084] Determine whether the node with the current minimum cost value is the node at the current tracking target position. If yes, generate an obstacle avoidance path based on the trajectory point set. If no, return to the step of determining the node with the minimum cost value from the node search set based on the cost value.

[0085] As described above, each iteration selects the node with the lowest value in the current generation for expansion, which helps to quickly find a better path from the current vehicle position to the target position, reduces the time required for path planning, meets the real-time requirements of autonomous vehicles, and the final obstacle avoidance path is obtained by considering multiple factors such as obstacles, vehicle kinematics, and path value, and has high quality. It can enable the vehicle to bypass obstacles in the best way possible while ensuring safety and continuing to travel along the tracking path.

[0086] Furthermore, calculating the cost value of each node in the node search set includes:

[0087] ;

[0088] ;

[0089] ;

[0090] In the formula, f(n) represents the cost of node n, g(n) represents the cost from the node at the vehicle location to node n, h(n) represents the expected value from node n to the node at the current tracking target location, g(n-1) represents the cost from the node at the vehicle location to the previous node n-1, and L n-1,n Let represent the distance traveled from the previous node (n-1) to node n, and addcosts represent the penalty factor cost. This represents the first weighting coefficient. This represents the second weighting coefficient. L represents the third weighting coefficient. E L represents the Euclidean distance from node n to the node at the current tracking target position, without considering environmental obstacles. D L represents the Durbins curve path length from node n to the node at the current tracking target position, without considering environmental obstacles. OThis represents the reciprocal of the nearest distance between node n and the obstacle.

[0091] As described above, the cost of each node consists of two parts: the historical cost (g(n)) and the expected cost (h(n)). To achieve a smoother path, a penalty factor, addcosts, is added to the historical cost calculation. The expected cost is calculated by a heuristic cost function, which consists of three parts. The weights of these three parts can be adjusted: increasing the weight of the third weight coefficient makes the search path as far away from obstacles as possible; increasing the weight of the first weight coefficient makes the search path biased towards the shortest path; and increasing the weight of the second weight coefficient makes the search path more consistent with the vehicle's kinematic trajectory, thereby ensuring the reliability of the obstacle avoidance path.

[0092] The obstacle avoidance method and system for unmanned vehicles described above are applicable to unmanned vehicles, and are explained below through specific embodiments:

[0093] Please refer to Figure 1 , Figure 3 and Figure 4 Embodiment 1 of the present invention is as follows:

[0094] A method for obstacle avoidance by an autonomous vehicle includes the following steps:

[0095] S1. Receive the prior static map of the autonomous vehicle and the tracking path on the prior static map.

[0096] S2. Based on the tracking path, determine the current tracking target position according to the driving information of the autonomous vehicle, such as... Figure 4 As shown, specifically including S21-S22:

[0097] The driving information includes the vehicle's location and driving direction.

[0098] S21. Obtain the pre-aiming distance of the unmanned vehicle.

[0099] S22. Determine the current tracking target position that is at the pre-aiming distance from the unmanned vehicle in the tracking path based on the vehicle position and the driving direction of the unmanned vehicle.

[0100] S3. Determine whether there is an obstacle in front of the unmanned vehicle. If yes, execute S31-S32. If no, do not perform obstacle avoidance planning, continue tracking to the current tracking target position, and return to execute S2.

[0101] In one alternative implementation, the presence of an obstacle ahead is detected by the millimeter-wave radar or lidar mounted on the autonomous vehicle; if an obstacle is detected, steps S31-S32 are executed.

[0102] like Figure 4 As shown, in another optional implementation, it is determined whether there is an obstacle in front of the autonomous vehicle. If there is, it is determined whether the distance between the current position of the autonomous vehicle and the previously planned position exceeds the length of the autonomous vehicle. If yes, S31-S32 are executed. If not, it is determined whether the previously planned path is not empty and whether there are no obstacles on the path. If it is not empty and there are no obstacles, obstacle avoidance planning is not performed, and the previously planned path is used. Otherwise, S31-S32 are executed. This avoids repetitive planning. If the vehicle movement distance is short and there are no obstacles intruding on the previously planned path, the previous planning result can be used again. This avoids planning uncertainty and instability due to slow vehicle movement, ensuring that the route remains stable within a certain movement distance. In addition, the computational workload of determining whether there are obstacles on the old path is smaller, while the computational workload of replanning is larger, thus reducing the computational workload.

[0103] S31. Based on the driving information and the current tracking target position, the prior static map is segmented to obtain a sub-map, and the obstacles are mapped onto the sub-map to obtain an obstacle distance information map, specifically including S311-S315:

[0104] S311. Determine the road segment to be traveled based on the vehicle position and the current tracking target position.

[0105] Specifically, the vehicle's position is taken as the starting point of the road segment to be traveled, and the current tracking target position is taken as the ending point of the road segment to be traveled, thus obtaining the road segment to be traveled.

[0106] S312. Determine the drivable area based on the road segment to be driven.

[0107] In one optional implementation, the vehicle position in the road segment to be driven is extended backward by one vehicle length, and the current tracking target position in the road segment to be driven is extended forward by one vehicle length to obtain a drivable area.

[0108] S313. Segment the drivable area from the prior static map to obtain a sub-map, such as... Figure 3 As shown.

[0109] S314. Discretize the sub-map to obtain a two-dimensional grid map, where each grid cell in the grid map represents a node, and the node represents a possible trajectory point to be traveled.

[0110] S315. Map the obstacles onto the grid map to obtain an obstacle distance information map.

[0111] By mapping obstacles onto a grid map, it is possible to determine the closest distance information between each node and the obstacle.

[0112] S32. Based on the obstacle distance information map, a hybrid A* path planning method is used to plan the path for the autonomous vehicle, resulting in an obstacle avoidance path, such as... Figure 4 As shown, specifically including S321-S328:

[0113] S321. Obtain an initial node search set from the obstacle distance information map, the node search set including nodes of the vehicle position.

[0114] S322. Calculate the cost value of each node in the node search set, and determine the node with the minimum cost value from the node search set based on the cost value.

[0115] The calculation of the cost value of each node in the node search set includes:

[0116] ;

[0117] ;

[0118] ;

[0119] In the formula, f(n) represents the cost of node n, g(n) represents the cost from the node at the vehicle's location to node n, i.e., the travel distance from the node at the vehicle's location to node n, i.e., the historical cost, h(n) represents the expected value from node n to the node at the current tracking target location, i.e., the estimated distance from node n to the node at the current tracking target location, i.e., the expected cost, used to guide the search direction, g(n-1) represents the cost from the node at the vehicle's location to the previous node n-1, L n-1,n This represents the distance traveled from the previous node (n-1) to node n. `addcosts` represents the penalty costs, including penalties for switching between forward and reverse driving, reversing, and having a different steering wheel angle than the previous node. This represents the first weighting factor, which biases the search path towards the shortest path. This represents the second weighting coefficient, which makes the search path more consistent with the vehicle's kinematic trajectory. L represents the third weighting factor, which ensures the search path stays as far away from obstacles as possible. E L represents the Euclidean distance from node n to the node at the current tracking target position, without considering environmental obstacles. D L represents the Dubins curve path length from node n to the node at the current tracking target position, without considering environmental obstacles. E and LD Unaffected by obstacles, but constrained by space and motion, L O This represents the reciprocal of the nearest distance between node n and the obstacle. Node n can be understood as the number of planning steps or grid steps. Starting from grid 0, the cost of surrounding grids is calculated, and the grid with the lowest cost is found. Node n calculates the cost of the current grid, which requires calling the cost of the previous (n-1) grids. Each step calculates the cost, and the cost is the grid selection number for each step. Alternatively, it can be understood as the planned grid distance to be reached, dividing the planning of a path into the selection of each grid, with each grid selection being a step.

[0120] S323. Add the node with the minimum cost value to the trajectory point set.

[0121] S324. Generate multiple extended nodes based on the nodes with the minimum cost value according to vehicle kinematics.

[0122] The node with the minimum cost value is the parent node, and the multiple extended nodes are the child nodes.

[0123] S325. For each of the extended nodes, determine whether the extended node is located in the area of ​​the obstacle or has been added to the trajectory point set. If yes, then no processing is performed on the extended node; otherwise, calculate the cost of the extended node.

[0124] The calculation of the cost value of the extended node is performed in the same way as the calculation of the cost value of each node in the node search set, and will not be repeated here.

[0125] S326. If the extended node has been added to the node search set, determine whether the calculated cost value of the extended node is less than the cost value of the extended node in the node search set. If it is less, update the cost value of the extended node in the node search set. If it is greater than or equal to the cost value, skip the extended node and search for other nodes. If no other nodes are found, the path planning fails.

[0126] In one alternative implementation, the index of the parent node is updated along with the value of the extended node in the node search set.

[0127] S327. If the extended node is not added to the node search set, then add the extended node to the node search set.

[0128] S328. Determine whether the node with the current minimum cost value is the node of the current tracking target position. If yes, generate an obstacle avoidance path based on the trajectory point set. If no, return to the step in S322 to determine the node with the minimum cost value from the node search set based on the cost value.

[0129] In one alternative implementation, the method further includes: after the autonomous vehicle has completed traveling along the obstacle avoidance path, returning to execute S2.

[0130] Please refer to Figure 2 Embodiment two of the present invention is as follows:

[0131] An unmanned vehicle obstacle avoidance system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the various steps of the unmanned vehicle obstacle avoidance method in Embodiment 1.

[0132] In summary, this invention provides an obstacle avoidance method and system for autonomous vehicles. Based on the tracking path and the driving information of the autonomous vehicle, the current tracking target position is determined. If an obstacle exists in front of the autonomous vehicle, a prior static map is segmented based on the driving information and the current tracking target position to obtain a sub-map. The obstacle is then mapped into the sub-map to obtain an obstacle distance information map. Based on the obstacle distance information map, a hybrid A-Star path planning method is used to plan the path for the autonomous vehicle, resulting in an obstacle avoidance path. This significantly improves the efficiency of path planning when the autonomous vehicle detects an obstacle, as it segments the prior static map into sub-maps and performs subsequent path planning within the sub-map area. The hybrid A-Star path planning method, based on the obstacle distance information map, allows for flexible obstacle avoidance path planning according to the specific location and distance information of the obstacle, enabling the vehicle to better adapt to complex and changing driving environments and react promptly to suddenly appearing obstacles, thereby improving the real-time performance of obstacle avoidance for autonomous vehicles. In addition, during obstacle avoidance planning, each iteration selects the node with the lowest value in the current generation for expansion, which helps to quickly find a better path from the current vehicle position to the target position, reducing the time required for path planning and meeting the real-time requirements of autonomous vehicles. Furthermore, the final obstacle avoidance path is obtained by considering multiple factors such as obstacles, vehicle kinematics, and path value, and has high quality. It can enable the vehicle to bypass obstacles in the most optimal way and continue to travel along the tracking path while ensuring safety.

[0133] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent modifications made based on the content of the present invention specification and drawings, or direct or indirect applications in related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for obstacle avoidance by an unmanned vehicle, characterized in that, Including the following steps: Receive a prior static map of the autonomous vehicle and the tracking path on the prior static map; The current target location is determined based on the tracking path and the driving information of the autonomous vehicle. Determine whether there is an obstacle in front of the autonomous vehicle. If so, segment the prior static map based on the driving information and the current tracking target position to obtain a sub-map, and map the obstacle into the sub-map to obtain an obstacle distance information map. Based on the obstacle distance information map, the hybrid A-star path planning method is used to plan the path of the unmanned vehicle to obtain the obstacle avoidance path. The driving information includes the vehicle's location and direction of travel; Determining the current target location based on the tracking path and the driving information of the autonomous vehicle includes: Obtain the pre-aiming distance of the unmanned vehicle; Based on the vehicle position and driving direction of the unmanned vehicle, the current tracking target position, which is at a pre-aiming distance from the unmanned vehicle, is determined in the tracking path; The segmentation of the prior static map based on the driving information and the current tracking target position to obtain sub-maps includes: The road segment to be traveled is determined based on the vehicle's location and the current tracking target's location; Based on the road segment to be traveled, determine the drivable area; The drivable area is segmented from the prior static map to obtain a sub-map.

2. The obstacle avoidance method for an unmanned vehicle according to claim 1, characterized in that, The step of mapping the obstacles onto the sub-map to obtain the obstacle distance information map includes: The sub-map is discretized to obtain a two-dimensional grid map, where each grid cell represents a node; The obstacles are mapped onto the grid map to obtain an obstacle distance information map; The obstacle avoidance path is planned by the autonomous vehicle using a hybrid A* path planning method based on the obstacle distance information map, and includes the following: An initial node search set is obtained from the obstacle distance information map, the node search set including nodes of the vehicle's location; Calculate the cost value of each node in the node search set, and determine the node with the minimum cost value from the node search set based on the cost value; Add the node with the minimum cost value to the trajectory point set; Based on vehicle kinematics, multiple extended nodes are generated according to the node with the minimum cost value; For each of the extended nodes, determine whether the extended node is located in the area of ​​the obstacle or has been added to the trajectory point set. If yes, the extended node is not processed. If no, the cost of the extended node is calculated. If the extended node has been added to the node search set, then determine whether the calculated cost value of the extended node is less than the cost value of the extended node in the node search set. If it is less, then update the cost value of the extended node in the node search set. If the extended node is not added to the node search set, then add the extended node to the node search set; Determine whether the node with the current minimum cost value is the node at the current tracking target position. If yes, generate an obstacle avoidance path based on the trajectory point set. If no, return to the step of determining the node with the minimum cost value from the node search set based on the cost value.

3. The obstacle avoidance method for an unmanned vehicle according to claim 2, characterized in that, The calculation of the cost value of each node in the node search set includes: ； ； ； In the formula, f(n) represents the cost of node n, g(n) represents the cost from the node at the vehicle location to node n, h(n) represents the expected value from node n to the node at the current tracking target location, g(n-1) represents the cost from the node at the vehicle location to the previous node n-1, and L n-1,n Let represent the distance traveled from the previous node (n-1) to node n, and addcosts represent the penalty factor cost. α This represents the first weighting coefficient. β This represents the second weighting coefficient. L represents the third weighting coefficient. E L represents the Euclidean distance from node n to the node at the current tracking target position, without considering environmental obstacles. D L represents the Durbins curve path length from node n to the node at the current tracking target position, without considering environmental obstacles. O This represents the reciprocal of the nearest distance between node n and the obstacle.

4. An obstacle avoidance system for an unmanned vehicle, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it performs the following steps: Receive a prior static map of the autonomous vehicle and the tracking path on the prior static map; The current target location is determined based on the tracking path and the driving information of the autonomous vehicle. Determine whether there is an obstacle in front of the autonomous vehicle. If so, segment the prior static map based on the driving information and the current tracking target position to obtain a sub-map, and map the obstacle into the sub-map to obtain an obstacle distance information map. Based on the obstacle distance information map, the hybrid A-star path planning method is used to plan the path of the unmanned vehicle to obtain the obstacle avoidance path. The driving information includes the vehicle's location and direction of travel; Determining the current target location based on the tracking path and the driving information of the autonomous vehicle includes: Obtain the pre-aiming distance of the unmanned vehicle; Based on the vehicle position and driving direction of the unmanned vehicle, the current tracking target position, which is at a pre-aiming distance from the unmanned vehicle, is determined in the tracking path; The segmentation of the prior static map based on the driving information and the current tracking target position to obtain sub-maps includes: The road segment to be traveled is determined based on the vehicle's location and the current tracking target's location; Based on the road segment to be traveled, determine the drivable area; The drivable area is segmented from the prior static map to obtain a sub-map.

5. The obstacle avoidance system for unmanned vehicles according to claim 4, characterized in that, The step of mapping the obstacles onto the sub-map to obtain the obstacle distance information map includes: The sub-map is discretized to obtain a two-dimensional grid map, where each grid cell represents a node; The obstacles are mapped onto the grid map to obtain an obstacle distance information map; The obstacle avoidance path is planned by the autonomous vehicle using a hybrid A* path planning method based on the obstacle distance information map, and includes the following: An initial node search set is obtained from the obstacle distance information map, the node search set including nodes of the vehicle's location; Calculate the cost value of each node in the node search set, and determine the node with the minimum cost value from the node search set based on the cost value; Add the node with the minimum cost value to the trajectory point set; Based on vehicle kinematics, multiple extended nodes are generated according to the node with the minimum cost value; For each of the extended nodes, determine whether the extended node is located in the area of ​​the obstacle or has been added to the trajectory point set. If yes, the extended node is not processed. If no, the cost of the extended node is calculated. If the extended node has been added to the node search set, then determine whether the calculated cost value of the extended node is less than the cost value of the extended node in the node search set. If it is less, then update the cost value of the extended node in the node search set. If the extended node is not added to the node search set, then add the extended node to the node search set; Determine whether the node with the current minimum cost value is the node at the current tracking target position. If yes, generate an obstacle avoidance path based on the trajectory point set. If no, return to the step of determining the node with the minimum cost value from the node search set based on the cost value.

6. The obstacle avoidance system for unmanned vehicles according to claim 5, characterized in that, The calculation of the cost value of each node in the node search set includes: ； ； ； In the formula, f(n) represents the cost of node n, g(n) represents the cost from the node at the vehicle location to node n, h(n) represents the expected value from node n to the node at the current tracking target location, g(n-1) represents the cost from the node at the vehicle location to the previous node n-1, and L n-1,n Let represent the distance traveled from the previous node (n-1) to node n, and addcosts represent the penalty factor cost. This represents the first weighting coefficient. This represents the second weighting coefficient. L represents the third weighting coefficient. E L represents the Euclidean distance from node n to the node at the current tracking target position, without considering environmental obstacles. D L represents the Durbins curve path length from node n to the node at the current tracking target position, without considering environmental obstacles. O This represents the reciprocal of the nearest distance between node n and the obstacle.

Images