Application of an Improved RRT Algorithm in Path Planning of Disordered Parcel Sorting Manipulator

Unordered parcel sorting corresponds to collision-free motion planning for manipulators within 3D unstructured obstacle-laden environments. Random poses of sorted objects impose stringent requirements on planning real-time performance and trajectory smoothness. The standard rapidly exploring random tree (RRT) relies on uniform random sampling and fixed-step expansion, which tends to yield weak sampling bias toward the goal, excessive invalid nodes, jagged redundant paths, and poor planning efficiency in complex environments under such three-dimensional constraints. To tackle these limitations, this paper proposes an improved algorithm based on the rapidly expanding random tree (RRT) algorithm and combining multiple strategies: GAB-RRT algorithm. In the sampling stage, the gravity bias strategy and the adaptive step size strategy are introduced to improve the nodes of the algorithm, expand the purpose and reduce invalid sampling. In the path planning link, the strategy of traversing connecting nodes to avoid obstacles and eliminate redundancy, and the strategy of cubic B-spline local optimization are combined to realize path simplification and smooth optimization. Considering the characteristics of the working environment, three-dimensional sorting space simulation experiments and robot path planning simulation experiments were designed and completed. The experimental results show that, compared with the standard RRT algorithm, the GAB-RRT algorithm is significantly optimized in terms of path cost, time cost and the number of nodes, among which the time cost is reduced by over 51%, the number of nodes is reduced by 87.3%, and the path cost is reduced by 4.93%, which can meet the requirements of high-precision path planning of robotic arms in the disorderly parcel sorting scene and provide a reference for the research on improving the performance of intelligent sorting equipment in the logistics sorting industry.