Computer Engineering and Applications ›› 2021, Vol. 57 ›› Issue (20): 157-163.DOI: 10.3778/j.issn.1002-8331.2012-0053

Previous Articles     Next Articles

Real-Time Sampling Path Replanning of Mobile Robot

TU Rui, WANG Wenge, LU Chengyang   

  1. College of Mechanical and Vehicle Engineering, Hunan University, Changsha 410082, China
  • Online:2021-10-15 Published:2021-10-21

移动机器人实时采样路径重规划

涂睿,王文格,卢成阳   

  1. 湖南大学 机械与运载工程学院,长沙 410082

Abstract:

A real-time sampling replanning algorithm DRT-RRT* using the reverse optimal rapidly-exploring random tree is proposed to solve poor path quality, serious jitter, poor real-time optimization effect when the traditional sampling-based planning algorithm is replanning in the dynamic environment due to its strong randomness. Firstly, a pruning strategy based on triangle inequality is introduced to smooth the path to reduce the inflection point. Then, in the process of optimizing the execution path, a combined sampling strategy and a local terminal jumping strategy are proposed, which focus the optimization target from the global path to the local path segment from the current position of the robot to the nearest inflection point, and then the execution path segment is modified in real time, so as to improve the stability of the path quality. Next, only the affected random trees are pruned in the path replanning, and the goal-biased sampling strategy is introduced when the random tree grows again, which works together with the combined sampling strategy to improve the search speed and stability of the path. Finally, RRT*-Pruning which is added triangular inequality pruning strategy and DRT-RRT* and RRT* are compared and analyzed. The experimental results verify the efficiency and stability of DRT-RRT* replanning.

Key words: mobile robot, path planning, Dynamic Real-Time Rapidly-exploring Random Trees*(DRT-RRT*), combined sampling, real-time path replaning

摘要:

针对传统采样规划算法因随机性强,在动态环境中重规划时路径质量差,抖动严重,实时优化效果不明显等问题,提出了一种利用反向生长最优快速搜索随机树的实时采样重规划算法DRT-RRT*(Dynamic Real-Time RRT*)。引入基于三角不等式的剪枝策略对路径进行平滑处理以减少路径拐点;提出了组合采样策略和局部终点跳动策略,将优化目标由全局路径聚焦于机器人当前位置至最近路径拐点的局部路径段,实时对执行路径段进行修正,进而提高路径质量的稳定性;在路径重规划时仅对受影响的随机树枝进行修剪,并在随机树重新生长时引入了目标偏置采样策略,与组合采样策略共同作用,提高路径搜索速率和稳定程度;将DRT-RRT*与RRT*和增加了三角不等式剪枝策略的RRT*-Pruning进行仿真对比分析,实验结果验证了DRT-RRT*重规划的高效性和稳定性。

关键词: 移动机器人, 路径规划, 动态实时-快速搜索随机树*(DRT-RRT*), 组合采样, 实时重规划