Parameter Identification of Non-Holonomic, Omnidirectional Vehicles using a Redundantly Parameterized Model
Sprache des Vortragstitels:
88th Annual Meeting of GAMM
Sprache des Tagungstitel:
Non-holonomic, omnidirectional vehicles have a high maneuverability and payload capacity but suffer from kinematic singularities. The usage of redundant coordinates allows to describe the systems dynamics without the need to switch between different sets of generalized coordinates. This paper presents a method for identifying the parameters of such a system. The method enhances a classic approach and makes it applicable to redundantly parametrized models and mobile systems. The method starts with formulating the equations of motion in a form linear w.r.t. a set of dynamic parameters. This equations contain, as a consequence of the redundant coordinates, unknown constraining forces. The latter are eliminated from the motion equations by application of an orthogonal complement to the constraint matrix. As the identification performance depends on the used trajectories special attention is paid to the motion planning. This is highly important for determining a set of uniquely identifiable basis parameters and the computation of a reliable identification trajectory. Reliable refers thereby to a motion which, on the one hand, optimally excites the parameters but, on the other hand, prevents the vehicle from losing traction (wheel slippage). The overall approach is validated by experiments and shown to yield promising results.