Experimental study of robotic polishing planning for uneven surfaces with force control

Abstract

Polishing technique plays a critical role in industrial manufacturing to smooth the surface of products such as turbine blades, die and mold. To improve the polishing process, many technologies are explored and applied. The automated polishing processes are categorized as computer numerically controlled machine polishing and industrial robotic polishing. The robotic arms possess the special advantages in working envelope, flexibility and production cost against computer numerically controlled machines. In addition, the robotic force control does improve the polishing performance in both surface material removal and surface roughness. This project strives to improve the polishing technologies by using combination of robotic techniques and intelligent computation. A specific polishing task that uneven surfaces should be polished to the required smoothness has been first defined. In order to solve the planning problem for this specific polishing task, the polishing process model is investigated and the polishing parameters are optimized based on computational intelligence. Neural network is employed to formulate the model while genetic algorithm is utilized to search for the best parameters. In experiments, workpiece surfaces under the robotic polishing using the planned parameters successfully achieve the desired polishing performance, which verifies the effectiveness of the proposed solution.