Adaptive control of deburring robots based on human skill models

A new controller for robotic deburring was developed on the basis of a human skill model. The architecture of this controller is a self-tuning adaptive control system. The gain tuning mechanism was designed to function just like the associative memories of a human expert, which enables the expert to perform skilful motion. The associative memories of a human expert were learned from teaching data acquired from the expert. A recursive least squares algorithm was applied to identify the process state. The stability analysis showed that the parameter estimation always converges to the true parameter values and the overall system is passive so that its response to any finite input is always stable. The experimental results of the system responses showed that the adaptive controller can adjust the reference feedrate in accordance with the process characteristics and that the adaptive controller can generate an effective reference feedrate with respect to the varying task conditions