A state model that incorporates robot arm dynamics together with joint
actuator dynamics is formulated as a nominally linear uncertain syste
m. The uncertainty is matched and it is shown to be bounded by a cubic
expression of the norm of the joint errors. A nonlinear robust feedba
ck controller is then derived which guarantees global ultimate bounded
ness of the actual system trajectories. Computer simulations are perfo
rmed on a model of the PUMA 560 robot arm and the results have verifie
d the feasibility of our controller with excellent trajectory tracking
performance.