| High speed robotic manipulators, such as direct drive manipulators, require effective controller designs. A number of adaptive control techniques have been proposed which offer stable and accurate performance. The direct and indirect methods of model-based adaptive control were selected for investigation.;The integration of a two degree-of-freedom, direct drive manipulator system is also described. Control algorithms for single joint motion were tested, and simulation results were validated.;Simulation of a two degree-of-freedom direct drive manipulator is used to demonstrate how adaptive control laws can lead to inferior performance when the richness condition is not satisfied. A non-identifiable desired trajectory is chosen and a realistic level of gaussian white noise is added to the velocity and acceleration measurements. Without persistent excitation, the control algorithms perform worse than a PD controller alone. Errors in parameter estimation, even after convergence, cause the manipulator to deviate from its desired path. Prior to convergence of the parameter estimates, adaptive control of the manipulator produces unacceptably large errors. |
