A New Auto-Tuned Velocity Estimator Design for Mechatronic Systems

Abstract

Velocity observers, specifically model independent observers, are of great value for controller implementation of mechatronic systems where the velocity of the system is either not measurable or the corresponding sensor is too costly. Therefore, the use of a surrogate signal is imperative. In this research, we present the design and associated stability of a saturation function based velocity estimator/observer for a general class of mechatronic systems. The proposed method estimates velocity without requiring prior knowledge of system dynamics. A novel adaptive update algorithm automatically adjusts the observer gains, eliminating the need for manual gain-tuning. We establish the stability of the velocity observer, along with the auto-tuned gain update rules, through a Liapunov-type analysis. Experimental validation on a rotary inverted pendulum demonstrates the effectiveness of the proposed method. Two case studies are conducted: the first compares the real-time performance of the proposed observer with a signum function based observer, while the second evaluates the observer's reliability with constant gains, highlighting its capability to support control design and maintain robustness against external disturbances.