Control and coordination of locomotion and manipulation of a wheeled mobile manipulator

In this thesis, we investigate modeling, control, and coordination of mobile manipulators. A mobile manipulator in this study consists of a robotic manipulator and a mobile platform, with the manipulator being mounted atop the mobile platform. A mobile manipulator combines the dextrous manipulation capability offered by fixed-base manipulators and the mobility offered by mobile platforms. While mobile manipulators offer a tremendous potential for flexible material handling and other tasks, at the same time they bring about a number of challenging issues rather than simply increase the structural complexity. First, combining a manipulator and a platform creates redundancy. Second, a wheeled mobile platform is subject to nonholonomic constraints. Third, there exists dynamic interaction between the manipulator and the mobile platform. Fourth, manipulators and mobile platforms have different bandwidths. Mobile platforms typically have slower dynamic response than manipulators. The objective of the thesis is to develop control algorithms that effectively coordinate manipulation and mobility of mobile manipulators. We begin with deriving the motion equations of mobile manipulators. The derivation presented here makes use of the existing motion equations of manipulators and mobile platforms, and introduces the velocity and acceleration dependent terms that account for the dynamic interaction between manipulators and mobile platforms. Since nonholonomic constraints play a critical role in control of mobile manipulators, we then study the control properties of nonholonomic dynamic systems, including feedback linearization and internal dynamics. Based on the newly proposed concept of preferred operating region, we develop a set of coordination algorithms for mobile manipulators. While the manipulator performs manipulation tasks, the mobile platform is controlled to always bring the configuration of the manipulator into a preferred operating region. The control algorithms for two types of tasks--dragging motion and following motion--are discussed in detail. The effects of the dynamic interaction are also investigated. To verify the efficacy of the coordination algorithms, we conduct numerical simulations with representative task trajectories. Additionally, the control algorithms for the dragging motion and following motion have been implemented on an experimental mobile manipulator. The results from the simulation and experiment are presented to support the proposed control algorithms