Path Planning for Robot Arms: Leveraging an Enhanced RRT-Connect Algorithm
Qun Zhao, Yan Shang, Yufan Zhou, XingWei Zhu, Rui Yan · 2025
In a complex three-dimensional environment, the path planning of a robot arm is the key to ensuring its efficient and safe completion of tasks. Traditional path planning algorithms such as Dijkstra and A* are challenging in meeting the requirements of real-time and high efficiency in high-dimensional complex space. Although the RRT-CONNECT (Rapidly-exploring Random Tree-Connect) algorithm shows high efficiency in path planning, its fixed step parameters lead to insufficient searching ability in narrow areas, and the generated path has redundancy and redirection problems. To solve these problems, this paper proposes a path-planning method for the robot arm based on an improved RRT-CONNECT algorithm. Firstly, we introduce the target bias strategy and optimize the sampling method to make the distribution of sampling points near the target point more natural and uniform and therefore improve the search efficiency. Secondly, the adaptive step size technology is used to subdivide the path segment and detect the collision point by point, which improves the accuracy of collision detection and the reliability of path planning. In addition, pruning technology and B-spline fitting technology are also introduced in this paper to remove redundant nodes and roundabout sections in the path, making the path smoother and more straightforward and further improving the robot arm’s path quality and navigation efficiency. The experimental results show that the improved algorithm can generate high-quality paths more efficiently in complex three-dimensional environments, significantly improve the path planning performance of the robot arm, and meet the real-time and high-efficiency requirements in practical applications.