Issues of modeling and control design for omnidirectional mobile robots with both structured and unstructured uncertainties are presented. First, system dynamics with an arbitrary location of the center of mass (CM) are derived. The system parameters, including the location of CM and those in the motor dynamics, are assumed unknown. The commonly adopted adaptive linearizing control suffers from the so-called control singularity under such circumstances. A smooth switching adaptive robust controller is then proposed to alleviate such a difficulty. It consists of four parts, a nominal adaptive linearizing controller, a deputy adaptive sliding-mode controller being in charge near singularity, a switching algorithm monitoring the exchange of control authority between the above two controllers, and a standard robust controller for handing the unstructured uncertainty. Not only does it significantly reduce the control voltages near singularity, but it also achieves better transients than a single adaptive controller and less control effort than a single robust controller. Both the simulation and experiments on an omnidirectional mobile robot are provided to demonstrate the validity of the proposed design.
關聯:
IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY 卷: 23 期: 5 頁碼: 1986-1993
