改进Bi-RRT*算法的自动泊车路径规划
Automatic Parking Path Planning with Improved Bi-RRT* Algorithm
姚智龙 1张小俊 1王金刚1
作者信息
- 1. 河北工业大学机械工程学院,天津 300132
- 折叠
摘要
针对双向快速搜索随机树算法(Bi-RRT*)生成的泊车路径不满足车辆运动学约束、路径曲折和收敛速度慢等问题,提出了一种改进Bi-RRT*算法.基于车辆碰撞检测模型,对障碍物进行膨胀处理,确保为车辆留出安全的泊车距离;采用Reeds-Shepp(RS)曲线对节点进行扩展,使路径连接满足车辆运动学约束;为了提高算法的搜索效率和采样成功率,引入了避障和径向约束采样策略;对路径节点进行均匀插值,利用基于起点的RRT*树对冗余路径进行剔除,实现路径平滑优化.仿真结果表明,与原始Bi-RRT*算法相比,改进后的Bi-RRT*算法规划出的路径不仅满足避障要求和运动学约束,而且规划时间和路径质量更具有优越性.
Abstract
To solve the problem that the parking path generated by the existing bidirectional rapidly-exploring random tree(Bi-RRT*)algorithmdoes not meet a vehicle's kinematics constraints,path twists and turns and slow convergence speed,an improved Bi-RRT* algorithm is proposed.First,based on the vehicle collision detection model,obstacles are magnified to ensure the safe parking distance for the vehicle.Secondly,the Reeds-Shepp curve is used to expand the nodes,so that the path connection meets the vehicle's kinematics constraints.Furthermore,in order to improve the search efficiency and sampling success rate of the Bi-RRT*algorithm,obstacle avoidance and radial constrained sampling strategies are introduced.Finally,the path nodes are uniformly interpolated,and the RRT*tree based on the starting point is used to remove redundant paths to realize the smooth optimization of the path.The simulation results show that compared with the existing Bi-RRT*algorithm,the path planned by the improved Bi-RRT*algorithm not only meets the obstacle avoidance requirements and kinematic constraints but is also superior in planning time and path quality.
关键词
自动泊车/路径规划/改进Bi-RRT*算法/Reeds-Shepp曲线/约束采样Key words
automatic parking/path planning/improved Bi-RRT*algorithm/Reeds-Shepp curve/constrained sampling引用本文复制引用
基金项目
天津市新一代人工智能科技重大专项(18ZXZNGX00230)
出版年
2024
国家科技期刊平台
NSTL
万方数据