A global path planning method for a low-speed autonomous driving scene
By improving the RRT* algorithm, combining dynamic step size adaptive adjustment, target bias adjustment, and region-restricted sampling, global path planning is optimized, solving the problem of low efficiency of the RRT* algorithm in low-speed autonomous driving scenarios, and achieving more efficient path search and shorter path length.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SOUTH CHINA UNIV OF TECH
- Filing Date
- 2022-11-22
- Publication Date
- 2026-07-21
AI Technical Summary
Existing RRT* algorithms cannot effectively improve the time efficiency of path search and optimize path length in low-speed autonomous driving scenarios, especially in complex environments where they cannot meet real-time requirements.
An improved RRT* algorithm is adopted, which optimizes global path planning through a dynamic step size adaptive adjustment strategy, a dynamic target bias adjustment method, a region-restricted sampling strategy, and a path smoothing module.
It significantly improves the time efficiency of path search, shortens the path length, and increases time efficiency by 75% and path length by 12%, meeting the real-time requirements of low-speed autonomous driving.
Smart Images

Figure CN115903815B_ABST
Abstract
Claims
1. A global path planning method for low-speed autonomous driving scenarios, characterized in that, It includes the following steps: Establish a grid map model of the entire working scenario, determine the positions of obstacles in the environment, and all passable paths; Set the starting point and the ending point; The global path planner searches for drivable routes based on the set start and end points. This global path planner employs an improved... The algorithm is used to search for driving paths, wherein the improved The improvements to the algorithm include a dynamic step-size adaptive adjustment strategy, a dynamic target bias adjustment method, and a region-restricted sampling strategy. The dynamic step-size adaptive adjustment strategy is based on x... new Node and starting point x init Euclidean distance between them, starting point x init With target point x goal The sampling step size is dynamically adjusted by the ratio of the Euclidean distance between them. The dynamic target bias adjustment method is that the bias probability at the end of the path search is greater than the bias probability at the beginning of the path search. During the movement, if an obstacle not marked on the map appears, enable the local path planning algorithm to re-plan the local path so that the vehicle can bypass the unknown obstacle; After bypassing the obstacle, use the global path planner again with the current point as the starting point and the target point as the ending point to plan the path and re-plan the path; The robot reaches the target point along the planned path; Among them, the dynamic target bias adjustment method is expressed as: If rat1 <= ratio1 and rand > ratio2, the random point is x_rand = [x_G, y_G]; If rat1 <= ratio1 and rand < ratio2, then the random point is x_rand = rand(1, 2) (x_L); If rat1 > ratio1 and rand < ratio3, the random point [x_rand(1), x_rand(2)] = Point_InCircle(x_new); If ratio > ratio1 and rand > ratio3, then for each node, x_rand = rand(1, 2). (x_L); Among them, rand is a random number between [0, 1], ratio1, ratio2, and ratio3 are all probability thresholds, x_G represents the abscissa of the target point; x_L represents the length of the environmental map, x_rand(1), x_rand(2) represent the abscissa and ordinate of the random sampling point, Point_InCircle(x_new) represents a function for sampling with the node x_new as the center, and rand(1, 2) represents a random function; The region-limited sampling strategy defines a sampling circular region with the newly generated node as the center and a preset length as the radius, thereby restricting the random point generation region.
2. The global path planning method for a low-speed autonomous driving scenario according to claim 1, characterized in that, According to x new Node and starting point x init Euclidean distance between them, starting point x init With target point x goal The sampling step size is dynamically adjusted based on the ratio of the Euclidean distances between the two paths. When the ratio of the distances is less than the distance threshold, the sampling step size changes according to the length of the generated path. When the ratio of the distances is greater than the distance threshold, it is converted to a fixed step size.
3. The global path planning method for a low-speed autonomous driving scenario according to claim 1, characterized in that, The dynamic step size adaptive adjustment strategy is expressed by the formula dis_sta2goal = dis_sta2new = rat1 = dis_sta2new / dis_sta2goal; If rat1 <ratio,Delta = step + step rat1; If rat1 ≥ ratio, Delta = step; Where dis_sta2goal represents the starting point x init With target point x goal The distance between them, where dis_sta2new represents the starting point x. init With x new The distance between nodes, x_G and y_G represent the x-coordinate and y-coordinate of the target point, respectively; ratio represents the distance threshold; and rat1 represents the starting point x. init Target point x goal The distance between them and the starting point x init x new Delta is the ratio of the distances between nodes, representing the sampling step size.
4. The global path planning method for a low-speed autonomous driving scenario according to claim 1, characterized in that, The improvement Improvements to the algorithm also include smoothing the path.
5. The global path planning method for a low-speed autonomous driving scenario according to claim 4, characterized in that, The smoothing of the path includes the steps: after generating the original path, check the connection of all points in the path. If there is no obstacle between the paths of the two connected points, delete all path points between these two points in the initial path and directly connect these two points; progress in turn, checking from the initial node to the target point.
6. The global path planning method for a low-speed autonomous driving scenario according to claim 4, characterized in that, After the smoothing of the path, use a path smoother to further optimize the path.
7. The global path planning method for a low-speed autonomous driving scenario according to claim 6, characterized in that, The path smoother uses the cubic B-spline interpolation method to further optimize the path.