Motion Primitive Based Predictive Control of a Rotational Actuator Driven by Contact and Friction

Abstract:
This work investigates the 2D-tilt control of a freely moving spherical cap, which is iteratively actuated only by contact and friction forces. To replace the extensive physical modeling and parameter identification effort of the complex actuation mechanism, a motion primitive based control strategy using experimental data is proposed. Based on predetermined input trajectories, a data-based model is identified, which represents the state-dependent steady-state transition when applying a motion primitive. A receding horizon optimal control strategy is employed to select the optimal motion primitive. Experimental results demonstrate the effectiveness of the proposed control approach under influence of disturbances.