Abstract: The proposed improved RRT-Connect algorithm (TRRT-Connect) addresses the issues of path elongation, excessive turns, and inadequate pass ability encountered in the standard RRT-Connect algorithm for path planning. Firstly, an improved RRT algorithm is employed to search and add a middle root node, facilitating the simultaneous expansion of four random trees to expedite algorithm convergence. Additionally, a target-biased strategy is employed for random point selection, and an attractive field is superimposed on node generation, along with integration of a greedy search strategy. Furthermore, a novel dynamic step size adjustment method is introduced, which dynamically selects appropriate step sizes by identifying the number of obstacles within the scanning region. Second, a bidirectional pruning optimization method is applied to the generated initial paths to accelerate pruning efficiency and remove redundant nodes along the paths. Finally, path smoothing is conducted at path turning points and reduce the number of paths turns. Simulation comparative experiments were conducted in three different environmental maps. The results indicate that the TRRT-Connect algorithm shows significant improvements compared to the standard RRT-Connect algorithm in terms of path length, number of iterations, and number of nodes. The paths generated are smoother without path turns, and there is better pass ability in densely populated obstacle areas. The experimental results confirm the effectiveness of this algorithm. Moreover, the application of the TRRT-Connect algorithm in field instance simulations reduces the transportation path length of mobile robots by 15.4% compared to traditional fixed paths, with smoother paths, further confirming the practicality of the algorithm.

Key words: RRT-Connect algorithm, dynamic step size adjustment, bidirectional pruning optimization, path planning

摘要： 针对标准RRT-Connect算法在路径规划中存在的路径冗长、转折较多和区域通过性欠缺问题，提出了一种新的改进RRT-Connect算法（TRRT-Connect）。首先，采用改进RRT算法搜索并添加一个中间根节点，实现同时扩展四棵随机树，加快算法收敛速度。并且在随机点的选取上使用目标偏置策略，在新节点的生成上叠加引力场，同时融合贪婪搜索策略。结合新的动态步长调节方法，通过识别扫描区域内障碍物的个数动态选择合适的步长。其次，对生成的初始路径使用双向剪枝优化方法，加快剪枝效率，剔除路径上的冗余节点。最后，对路径转折处进行光滑处理，减少路径转折。在三种不同环境地图中进行仿真对比实验，结果表明：TRRT-Connect算法与标准RRT-Connect算法相比较，在路径长度、迭代次数和节点数上有较大改善，在密集障碍物区域的通过性较好，路径更加光滑且无转折，实验结果证明了该算法的有效性。同时将TRRT-Connect算法应用于现场实例仿真中，使得移动机器人的运输路径长度相较于传统固定路径降低了15.4%，且路径光滑，进一步验证了该算法的实用性。

关键词: RRT-Connect算法, 动态步长调节, 双向剪枝优化, 路径规划