【作者】 李犇； 褚伟；

【Author】 LI Ben;CHU Wei;School of management,Hefei University of technology;

【机构】 合肥工业大学管理学院；

【摘要】 快速扩展随机树算法(Rapidly-Expanding Random Trees,RRT)是解决机器人路径规划问题的算法之一,但该算法搜索效率低、路径不平滑、复杂环境中无法对动态障碍物进行避障。针对以上问题,首先在RRT算法基础上提出了一种结合先验知识采样的快速扩展随机树算法(PrioriKnowledge-RRT,PK-RRT)。将已成功规划路径的关键节点作为先验知识构建数据库,在RRT采样过程中增加对数据库的概率采样,优化采样效率,重新布线优化路径平滑度。再提出基于PK-RRT与人工势场法(Artificial Potential Field Method,APF)的融合算法实现复杂环境下的动态避障。实验结果表明,PK-RRT算法在全局规划的时间、路径长度以及平滑度上均优于RRT算法;融合算法可以在复杂环境下安全、有效的完成机器人的动态路径规划。更多还原

【Abstract】 Rapidly-Expanding Random Trees(RRT) is one of the algorithms to solve the robot path planning problem,but this algorithm has low search efficiency,unsmooth paths,and cannot avoid dynamic obstacles in complex environments.To solve the above problems,first of all,based on RRT algorithm,an RRT algorithm combined with prior knowledge sampling(Priori Knowledge-RRT,PK-RRT) is proposed.The key nodes of the successfully planned path are used as prior knowledge to construct the database,and the probability sampling of the database is increased during the RRT sampling process,to optimize the sampling efficiency,and reroute to optimize the smoothness of the path.Then,a fusion algorithm based on PK-RRT and Artificial Potential Field Method(APF) is proposed to realize dynamic obstacle avoidance in complex environments.The experimental results show that the PK-RRT algorithm is superior to the RRT algorithm in the time,path length and smoothness of the global planning.The fusion algorithm can safely and effectively complete the dynamic path planning of the robot in a complex environment.更多还原