A group of physically meaningful, practical, and adaptive proportional-integral-derivative control robust against significant changes in robot and external dynamics

Abstract

This thesis proposes a physically meaningful, practical, and adaptive Proportional-Integral-Derivative (PID) control robust against significant changes in external and robot dynamics. The proposed PID control achieved robustness against significant inertia change and external disturbance force with Nussbaum based adaptive control law. Considering the practical implementation problem in a digital platform caused by measurement, quantization, and numerical calculation noise, the forgetting factor is applied in the proposed PID control. The forgetting factor of the proposed control law prevents incorrect gain calculation and gain drift by measurement, quantization, and numerical calculation noise that cannot represent the current system state. Furthermore, thanks to Time-Delay Estimation (TDE), the proposed PID control practically can be implemented in a digital system as it does not require an accurate robot model and complicated dynamics calculation. Besides, there are only two meaningful tuning parameters that are tuned by trial-and-error. The robustness against a significant robot and external dynamics variation verified with the simulation and experimental studies with 2 Degree-Of-Freedom (DOF) Direct-Drive planner robot. The simulation results showed that the proposed PID control robust against a significant robot and external dynamics changes. The experiment results showed that the proposed PID control achieves high robustness against significant robot dynamics changes, compared to the constant PID control.