Neural network-based tracking control for robotic systems using only position feedback
An adaptive neural network-based position feedback tracking control scheme for robotic systems involving plant uncertainties and external disturbances is proposed. The developed controller is based on a neural network system and a linear reduced-order observer. The resulting closed-loop system guarantees a transient and asymptotic performance, in the sense that the tracking error locally converges to a small region around zero in terms of L∞ bound and H∞ performance. The implementation of the neural network basis functions depends only on the desired reference information. Only position measurements are required for feedback, and the developed controller is driven by the position tracking error. Consequently, the adaptive neural network-based controller developed possesses the properties of computational simplicity and easy implementation. Finally, a simulation example is provided to illustrate the tracking performance of a two-link robotic manipulator.
