arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2305.08080cs.RO
收藏

概率RRT Connect与中间目标点选择用于自动驾驶车辆在线规划

Probabilistic RRT Connect with intermediate goal selection for online planning of autonomous vehicles

Virginia Tech(弗吉尼亚理工大学)

Darshit Patel, Azim Eskandarian

AI总结:

本文提出一种结合概率RRT与RRT-Connect的改进算法,通过选取中间目标点构建多棵树,快速生成绕障安全路径,并用MPC+Stanley控制器跟踪,轨迹更优且符合人类直觉。

摘要:

快速探索随机树(RRT)是机器人领域运动规划中应用最广泛的算法之一。为减少探索时间,引入了RRT-Connect算法,该算法同时构建两棵树并最终将其连接。概率RRT(pRRT)利用位置概率图的概念引入目标偏向,以加快收敛速度。本文提出一种改进方法,将pRRT与RRT-Connect技术相结合,以快速获得绕开障碍物的可行轨迹。不同于从起点到终点构建单棵树,我们在障碍物周围选择中间目标点,构建多棵树以连接起点、终点及中间目标点,这些局部树最终连接形成绕开障碍物的整体安全路径。所获路径由MPC+Stanley控制器跟踪,生成每一步带有控制指令的轨迹。所提方法生成的轨迹更优且符合人类直觉。将该算法与标准RRT和pRRT进行比较,以研究其相对性能。

英文摘要:

Rapidly Exploring Random Trees (RRT) is one of the most widely used algorithms for motion planning in the field of robotics. To reduce the exploration time, RRT-Connect was introduced where two trees are simultaneously formed and eventually connected. Probabilistic RRT used the concept of position probability map to introduce goal biasing for faster convergence. In this paper, we propose a modified method to combine the pRRT and RRT-Connect techniques and obtain a feasible trajectory around the obstacles quickly. Instead of forming a single tree from the start point to the destination point, intermediate goal points are selected around the obstacles. Multiple trees are formed to connect the start, destination, and intermediate goal points. These partial trees are eventually connected to form an overall safe path around the obstacles. The obtained path is tracked using an MPC + Stanley controller which results in a trajectory with control commands at each time step. The trajectories generated by the proposed methods are more optimal and in accordance with human intuition. The algorithm is compared with the standard RRT and pRRT for studying its relative performance.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑