基于改进RRT-Connect算法的移动机器人路径规划

Path Planning for Mobile Robot Based on Improved RRT-Connect Algorithm

针对双向快速扩展随机树(RRT-Connect)算法在路径规划过程中采样随机性大、目标导向差、搜索效率低等问题,提出了一种基于区域约束采样的RRT-Connect算法。首先,该算法在进行节点扩展的过程中,将采样点限制在由对方搜索树末端所形成的约束区域,提高目标导向性,同时根据该区域形成反向次采样区域,用于逃脱局部震荡;其次,构建变转向角策略,减少随机采样过程中出现的逆生长现象;最后,利用贪婪策略消除路径冗余点以及二阶贝塞尔曲线进行路径平滑。通过多组仿真实验表明,改进RRT-Connect算法在平均路径长度、平均规划时间、有效节点利用率上均有较大改善,证明了改进机制的有效性。

To solve the problems of high sampling randomness,poor target orientation,and low search efficiency in the path planning process of the bidirectional fast expanding random tree(RRT-Connect)algorithm,a region constrained sampling based RRT-Connect algorithm is proposed.Firstly,in the process of node expansion,the algorithm restricts the sampling points to the constraint area formed by the end of the other side′s search tree,to improve target orientation.At the same time,a reverse sub-sampling area is formed based on the area to escape local oscillations;Secondly,a variable steering angle strategy is constructed to reduce the reverse growth phenomenon that occurs during the random sampling process;Finally,greedy strategy is used to eliminate redundant points in the path and second-order Bessel curve is used for path smoothing.Multiple simulation experiments have shown that the improved RRT-Connect algorithm has significant improvements in average path length,average planning time,and effective node utilization,which demonstrates the effectiveness of the improved mechanism.