Model Predictive Force Control in Grinding based on a Lightweight Robot

摘要

The extensive effort in manual labour for mould manufacturing is the main reason for outsourcing manufacturing resources to low-wage countries. Furthermore, the international mould and die sector faces a lack of skilled workers due to the unattractive working conditions, for example the stressing, monotonous work and a partly hazardous working atmosphere (e.g. nickel dust). Especially, the finishing process (e.g. grinding, lapping, polishing) of freeform surfaces is still less automated in metal processing. Therefore, many efforts are initiated in order to fully automate the finishing process, but the today's approaches are usually limited on prototypal implemented and cost-intensive machine solutions. This work introduces a finishing control approach for freeform surfaces using a lightweight robot. The robot is used as an actuator as well as a sensor for closed-loop force control. The optimal force trajectory is determined by a CAD-model which describes the geometry of the workpiece. The noisy measured force signal is filtered by a Kalman filter which also estimates the system state for the model predictive controller (MPC). Finally, a simple position controller controls the position of the tool center point (TCP) and the MPC controls the force which occur on the tool. The results show that the force can be controlled with a higher accuracy than with a conventional impedance control, which is commercially provided.