Self-tuning Control for Articulated Robots Using the Plestan's Method

Abstract

A self-tuning controller is proposed for an articulated robot using the Plestan’s method. To this end, we reconstruct the articulated robot dynamics exploiting the time-delay estimation (TDE) technique. The closed-loop error dynamics is described with sliding variables and TDE error; then, the Plestan’s sliding mode based gain-adaptation law is incorporated with the TDE technique. The stability of the overall dynamics is proven in the sense of Lyapunov. As a result, self-tuning of the gain is realized through the sliding variable. When the TDE error increases due to the nonlinear effect such as friction, the adaptive gain is automatically adjusted to counteract the TDE error. Chattering can be avoided because the sliding mode based gain dynamics does not allow the gain increase to an excessively high value. The superiority of the proposed self-tuning controller is demonstrated by comparative experiments on a multiple joints robot setup.