On robust H∞-based control design for flexible joint robots

Control design of robot manipulators requires a nominal model. This model may describe the physical system correctly, but it is only an approximation. Several types of uncertainty are prevalent and must be considered in control system design. The uncertainty may be due to unmodeled dynamics, neglected nonlinearities, compliance effect of gear transmission mechanism in the joint, measurement noise, or environmental uncertainty. Therefore, control design methods must account for model inaccuracies or uncertainties. This work concerns the improvement of the performance of flexible joint robots with harmonic drives as gear train mechanism. A synthesis technique based on H∞ for robots with uncertain parameters that include the actuator model is utilized. H_∞ design framework has the advantages of allowing us to design for performance and robustness simultaneously. In this method the error model of the rigid body dynamics is derived. Then, a high-level H∞-based controller is designed for the error dynamics. Finally, a low-level controller based on PID to compensate for the effects of joint flexibility and friction is designed. To validate the proposed method, a one DOF flexible joint robot has been considered for simulation. Consecutively to evaluate the robustness property of the proposed control strategy, the payload or the configuration of the robot will be changed. Control performance will then be evaluated by changing the link inertia