改进双向Informed-RRT*算法的室内移动机器人路径规划研究

Research on Path Planning for Indoor Mobile Robots Based on Improved Bidirectional Informed-RRT* Algorithm

摘要:

针对移动机器人在室内环境下存在的路径收敛速度慢、效率低、路径轨迹离障碍物近且平滑度欠佳等问题，提出了一种改进双向Informed-RRT*算法。该方法通过引入两棵生长树，对采样空间进行交替相向的搜索，同时以一定概率进行随机点的目标偏置选择，使得更快获得初始路径；根据当前节点周围环境进行自适应步长扩展，在碰撞低风险区获得更长的路径步长；对新节点进行父节点重选和重布线操作，优化路径质量；修剪初始路径的冗余节点，缩短算法采样空间，减少无效采样；为保证移动机器人安全行走，设置障碍物层，并采用局部分段三次Hermite插值平滑最终路径。在Matlab和ROS机器人仿真平台分别进行2D和3D的实验，仿真结果表明，改进算法在路径规划时间、路径节点数及路径长度上更具可行性和优越性，且在含动态障碍物的室内场景下表现出更强的鲁棒性。

Abstract:

To address the problems of slow path convergence, low efficiency, and poor smoothness of paths close to obstacles in indoor environments for mobile robots, an improved bidirectional Informed-RRT* algorithm is proposed. This method introduces two growing trees to perform alternating, opposing searches of the sampling space, while simultaneously selecting target biases for random points with a certain probability, thus achieving faster initial path acquisition. It adaptively expands the step size based on the surrounding environment of the current node, obtaining longer path step sizes in low-collision-risk areas. Parent node reselection and rewiring operations are performed on new nodes to optimize path quality. Redundant nodes in the initial path are pruned to shorten the algorithm's sampling space and reduce invalid sampling. To ensure safe movement of the mobile robot, an obstacle layer is set up, and local segmented cubic Hermite interpolation is used to smooth the final path. Experiments were conducted in 2D and 3D on Matlab and ROS robot simulation platforms, respectively. Simulation results show that the improved algorithm is more feasible and superior in terms of path planning time, number of path nodes, and path length, and exhibits stronger robustness in indoor scenarios with dynamic obstacles.