Swept Volume SDF-Based Feeding and Unloading Path Planning Method for Bent Workpieces

Abstract

In robot-assisted bending operations, the confined space between the upper and lower dies of the machine, combined with the dynamic shape alterations of the workpiece, presents a significant risk of collisions. Conventional Rapidly-exploring Random Tree (RRT) algorithms exhibit limitations in terms of slow convergence and suboptimal path quality under such conditions. Additionally, many existing approaches prioritize path efficiency without adequately considering path safety. To address these limitations, a hierarchical path planning method is proposed. First, we use an enhanced RRT algorithm, called improved synchronized-biased greedy RRT-Connect (SBG-RRT-Connect), to generate initial paths more efficiently. This algorithm improves convergence and avoids unproductive areas, speeding up the process and reducing search time by 67.05% and iterations by 76.06%. Next, we introduce a safety measure based on the Swept Volume Signed Distance Field (SVSDF) of the workpiece’s motion, which helps identify potential collision risks. Multi-objective optimization is achieved using the Water Cycle Algorithm (WCA), enabling a balance between path cost and safe distance from obstacles. Experimental results verify the method’s applicability to workpieces with varying complexities. The path effectiveness is further validated through a digital twin-based simulation platform and physical implementation.