Dynamic Model-Based Control of Redundantly Actuated, Non-Holonomnic, Omnidirectional Vehicles

Vehicles with several centered orientable wheels have one of the highest maneuverability and are hence an excellent choice for transportation tasks in narrow environments. However, they are non-holonomic, in general redundantly actuated, and additionally sufferfromconfiguration singularities, which makes their modeling and control challenging. Existing control approaches only consider the vehicle kinematics whereas the required torques are commonly controlled by classical PD motor controllers. However, this leads to considerable tracking errors and a violation of the constraints especially during acceleration phases. Moreover, actuator counteractions and an undefined torque distribution can be observed. This paper introduces a model-based control concept that overcomes these issues. It resolves counteractions and distributes torques according to physical limitations which significantly reduces slippage and the energy consumption and further reduces the tracking error. To this end, an inverse dynamics solution of a redundantly parametrized model is used. The method is robust to configuration singularities. This is confirmed by experimental results.