Uncertainty-aware error modeling and hierarchical redundancy optimization for robotic surface machining

Abstract

Industrial robots are commonly utilized in in-situ machining, but their inherent limitations in stiffness and positioning accuracy can pose challenges in achieving precise freeform surface milling. Previous research primarily focused on robot stiffness optimization and positioning error prediction separately, neglecting the uncertainty of robot errors and the modeling of surface profile errors. To address these issues, this work introduces a novel evaluation method for surface profile errors in robotic milling. It combines geometric error prediction models, based on kinematic and stiffness models, with non-geometric errors using Gaussian process regression. Moreover, collaborative optimization of multiple redundancies in robotic surface milling systems is crucial for reducing machining errors effectively. To achieve this, this work proposes a collaborative optimization method for optimizing the robot’s posture change sequence and the workpiece’s orientation, considering the motion constraints and error requirements of the robot. Among them, this work introduces the improved Whale Optimization Star Algorithm (WOA*) method to address the optimization problem of coupling robot posture and workpiece’s orientation. The proposed method’s effectiveness is confirmed through simulations and experiments, which clearly demonstrate its capability in improving error prediction and machining accuracy in robotic milling.