An Improved RRT* Algorithm Based on Collision Density Detection for Manipulator Path Planning

To address the limitations of the RRT* algorithm in three-dimensional path planning, including insufficient obstacle-overcoming ability, susceptibility to local optima, and low search efficiency, this paper proposes an improved algorithm. The proposed method combines a heuristic cost function with an adaptive adjustment mechanism, leveraging local collision density detection to dynamically adjust the gain factor k. This optimizes the distribution of sampling points, enabling efficient obstacle traversal, escaping from local optima, and improved planning speed while maintaining path quality. Simulation experiments were conducted to compare the performance of the RRT algorithm, standard RRT* algorithm, heuristic-enhanced RRT* algorithm, and the proposed algorithm in three different environments. The results demonstrate that the proposed approach accelerates convergence, significantly enhances obstacle-overcoming and local optima avoidance capabilities, and ensures stable path quality.