An adaptive state observer for rigid link manipulators is proposed. Th
e proposed adaptive state observer is based on parametrizing the robot
dynamic model via the estimated state and parameters, thus yielding a
nominal model plus two perturbation terms, one is linear and the othe
r is quadratic in the state and parameter estimation errors. The updat
e law for the adaptive observer is derived using the MKY lemma (Narend
ra and Annaswamy 1989). Stability of the origin of the overall error s
ystem is shown using the Gronwall-Belmann lemma (Khalil 1992) and a co
nverse Lyapunov theorem (Khalil 1992). The proposed adaptive observer
is tested through simulations.