Obstacle Avoidance Path Planning for a Basket-Carrying Robotic Arm Based on an Improved APF-RRT Algorithm

Abstract

This study addresses the challenges of robotic arm path planning in unstructured material-handling environments typical of agricultural robotics, including complex obstacle distributions, weak goal-directedness of conventional RRT algorithms, low convergence efficiency, and poor path smoothness. An RRT planner embedding an improved artificial potential field (APF) is proposed. A power law decay repulsive gain removes goal unreachability; a three-level progressive escape mechanism with a penalty-weighted distance metric bypasses local deadlock regions; environment-aware adaptive step size and a dynamic goal bias probability balance global exploration with local convergence. Multi-pass greedy pruning and cubic B-spline smoothing further remove redundancy and ensure a smoother path. Comparative simulations and real-robot grasping experiments were conducted in 2D and 3D configuration spaces and on a 6-DOF collaborative robotic arm platform. In 2D scenarios, the improved algorithm reduces average iteration counts by 85.8%–89.7% and planning time by 42.1%–56.4% relative to conventional RRT. In 3D space, planning is completed in an average of 17.1 iterations with path length shortened by 31.4%. In joint space experiments with the 6-DOF robotic arm, the algorithm reduced path cost by 38.9% and 43.9% compared to RRT and RRT-Connect, respectively, and shortened planning time by 10.1%–20.9%. Physical experiments demonstrated controlled tracking errors and smooth motion. Real-robot trials confirm smooth, jitter-free motion across all joints, demonstrating that the algorithm effectively satisfies engineering requirements for path quality and motion stability in complex agricultural environments.