Research on a Path Planning Algorithm for Leather-Grasping Robots Based on Improved RRT

Abstract

To address the randomness, inefficiency, and lack of guidance inherent in traditional Rapidly-exploring Random Tree (RRT) algorithms when applied to the path planning of dual-arm robots for spreading and grasping leather, an improved RRT algorithm is proposed. Considering the flexible nature of leather, the concept of a dynamic artificial potential field is introduced to dynamically adjust attractive and repulsive forces, guiding the random tree’s growth toward the target direction. This approach reduces blind searches, enhances the algorithm’s guiding capability, and accelerates planning. Furthermore, cubic non-uniform B spline curves are employed to smooth the planned path, ensuring a continuous trajectory for the robotic end-effector. Simulation results demonstrate that the improved algorithm achieves higher search efficiency and lower path costs when planning the grasping path for leather corners. Compared with the traditional RRT algorithm, the proposed method generates superior paths for leather manipulation tasks