非完整约束移动机器人CRAB RRT轨迹规划  

CRAB-RRT Trajectory Planning Algorithm for Nonholonomic Constraints Mobile Robots

作　　者：宋云云 杨盛毅[2,3] 朱力 SONG Yunyun;YANG Shengyi;ZHU Li(School of Data Science and Information Engineering,Guizhou Minzu University,Guiyang 550025,China;Key Laboratory of Pattern Recognition and Intelligent System,Guizhou Minzu University,Guiyang 550025,China;School of Mechatronics Engineering,Guizhou Minzu University,Guiyang 550025,China)

机构地区：[1]贵州民族大学数据科学与信息工程学院,贵州贵阳550025 [2]贵州民族大学模式识别与智能系统重点实验室,贵州贵阳550025 [3]贵州民族大学机械电子工程学院,贵州贵阳550025

出　　处：《机械与电子》2023年第4期3-8,16,共7页Machinery & Electronics

基　　金：贵州省科学技术基金项目(黔科合基础[2017]1088)。

摘　　要：针对快速扩展树算法在多障碍环境搜索效率低、冗余节点多的问题,提出一种CRAB-RRT算法。该算法采用圆域概率偏向采样策略进行随机采样;将扩展树上节点分为2类,通过度量邻近点间对角线距离、历史路径成本和扩展角度的加权和,在有效邻近点集中选出高质量最近点;在新节点扩展中添加了目标引力分量,并根据环境障碍信息实现扩展步长自适应变化;经过对非完整约束移动机器人运动学模型的分析,提出了极限角约束,通过考虑受极限角约束的路径裁剪策略剔除冗余节点,利用三次B样条平滑方法平滑轨迹。通过数值仿真分析和实车实验,验证了CRAB-RRT算法的可行性、实用性与优越性,平滑后的路径更利于机器人跟踪,且适用于多障碍环境。Aiming at the problem of low search efficiency and many redundant nodes of the rapidly-exploring random trees algorithm in multi-obstacle environment,the CRAB-RRT algorithm is proposed.In this paper,random sampling can be implemented by the circle domain probability bias sampling strategy.Nodes in the extension tree are divided into two classes,and the nearest points with high quality are selected from set of efficient neighboring points by comparing the weighted sum of the diagonal distance between adjacent points,the historical path cost and the extension angle.The target gravity component is added in the new node expansion,and the expansion step adaptively changes is related to the information of environmental obstacles.After analyzing the kinematic model of nonholonomic mobile robot,the limit angle constraint was proposed.The redundant nodes were eliminated by path clipping strategy with limit angle constraint,and the trajectory was smoothed by cubic B-spline smoothing method.The feasibility and practicability of the CRAB-RRT algorithm are verified by numerical simulation analysis and real vehicle experiment.The smoothed path is more conducive to robot tracking and is suitable for multi-obstacle environment.

关 键 词：非完整约束 CRAB-RRT 目标偏向性 自适应步长 B样条平滑 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]