Observers for rigid robots

Abstract

In virtually all present-day robotic applications, the design approach is based on controllers which utilize measurements of both position and velocity. However, velocity measurements are often contaminated by noise, and cause the controller's performance to deteriorate. Therefore the function of the observer is essential for reconstructing the velocity signals. The present study establishes a new systematic procedure that creates the possibility of obtaining an observer for estimating the links velocity in a rigid robot. The underlying approach adopts the concept of semiglobal stabilizability, and the resulting observer's output converges asymptotically to the system's velocity vector. In particular, the present approach can be implemented to the case where the robot velocity is either not uniformly bounded or its upper bound is unknown.