Abstract
In this study, an adaptive force-position-speed collaborative process planning framework for robot polishing was proposed to improve the stability of the robot polishing process. The material removal model based on Preston's theory was studied, and the factors of polishing pressure, tool speed, feed speed, and sandpaper type were considered to design the manual polishing experiment. The improved Dung Beetle Optimization algorithm, Back Propagation Neural Network, Finite Element Analysis, and Response Surface Methodology provide a strong guarantee for the selection of robot polishing process parameters. For curved workpieces, the curvature adaptive interpolation method is introduced to generate trajectories. An adaptive impedance control strategy is implemented to enhance force control, and PD iteration and RBF neural networks are used to ensure stable contact force and accuracy. The experimental results show that the root mean square error (RMSE) accuracy of the established roughness prediction model reaches 0.0001 µm, the proposed force control method is more stable, and the surface roughness is reduced by 20.79% on average compared to the baseline method, which proves the effectiveness of the framework in achieving high precision and high efficiency of robot polishing.