多窄路口下改进的bi-RRT路径规划  

Improved bi-RRT Path Planning at Multiple Narrow Intersections

作　　者：龙建全 张皓然 梁艳阳[1] LONG Jian-quan;ZHANG Hao-ran;LIANG Yan-yang(Department of Information Technology, Southwest University of Science and Technology, Mianyang 621010, China)

机构地区：[1]西南科技大学信息工程学院

出　　处：《仪表技术与传感器》2019年第5期118-123,共6页Instrument Technique and Sensor

摘　　要：多窄路口的复杂环境路径规划中,快速扩展随机树(rapidly exploring random Trees,RRT)存在重复搜索和难以通过等缺点。提出改进的双向快速随机扩展树(bi-directional RRT,bi-RRT)的路径规划算法,在多路口来设置人工虚拟目标点,首先根据其连通域采用Dijsktra算法求出一组最短路径的虚拟目标点,再根据虚拟目标点构建采样区域,结合小车的非完整积分约束、环境约束和上述构建的采样区域,利用bi-RRT搜索可行路径。该算法解决了机器人在狭窄路口重复搜索的问题并提高全局搜索效率。通过仿真实验验证该算法的高效性、实时性和正确性。Path planning of multiple narrow intersections in the complex environment, rapidly exploring random trees (RRT) has the shortcomings of repeated search and hard to pass. An improved bi directional RRT (bi RRT) path planning algo rithm was proposed to set artificial virtual target points at multiple intersections. Firstly, according to the connected domain, Di jsktra algorithm was used to solve a set of shortest path virtual target points. Then a sampling region was constructed according to the virtual target points. Combined with the nonholonomic integral constraints and environmental constraints of the trolley, the bi -RRT search was feasible with the above sampling region to get path. The algorithm solved the problem of robot repeated search at a narrow intersection and improved the global search efficiency. The effectiveness, real time and correctness of the algorithm are verified by simulation experiments.

关 键 词：路径规划 多窄路口 快速扩展随机树 非完整约束 虚拟目标点 搜索效率 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]