Abstract
A classical state observer-based approach to velocity estimation for the position control of a robot manipulator is compared with a self-sensing approach. The former is essentially an analog differentiator, while the latter utilizes a model of the actuator to reconstruct the velocity signal. The two methods of velocity estimation are compared in two common robot control schemes implemented on a compliant-mechanism-based small-scale robot manipulator. Specifically, both the state-observer-based and the self-sensing velocity estimates are utilized in the PD-control and in the sliding-mode control of the manipulator motion. Experimental results indicate that the self-sensing approach offers better control performance than does the state-observer-based approach, especially in the sliding-mode controller. The results are particularly useful in the motion control of micromanipulators, since the small scale constrains the size and number of sensors on the manipulator.
