| Summary: | 碩士 === 國立中興大學 === 電機工程學系 === 91 === Abstract
This thesis develops methodologies and design techniques for modeling and tracking control of a wheeled mobile manipulator (WMM). This mobile manipulator system is typically composed of two rigid manipulators with three links mounted on a wheeled mobile robot. Compared to mobile robots and traditionally fix-based manipulators, this system is able to perform versatile missions in various applications. Owing to the wheeled structure, the nonholonomic constraints are introduced into system models to limit lateral mobility, thereby resulting in infeasibility of the smooth static feedback control laws. Meanwhile, exact modeling and controller synthesis of the WMM are nontrivial due to not only the complex system structure, but also coupled dynamics between mobile platform and manipulators.
The exact dynamic models and constraint equations of the vehicle are derived based on the Lagrange’s equation with multiplier and the commercial symbolic software packages. In order to understand the effects of the coupled dynamics, the system is decomposed into three subsystems. Structural properties (e.g., skew-symmetry) of the WMM are well analyzed, which are very useful in synthesizing tracking controllers for the type of vehicle. To insure the correctness of the proposed models, the models can be reduced to match some well-known models.
Using the backstepping technique and the filter-error method, the globally asymptotically stable tracking controller based on the Lyapunov theory is proposed to make the vehicle follow any smooth desirable trajectories. This control law is capable of dealing with nonholonomic constraints and fully dynamic interactions between mobile vehicle and manipulators.
Several path following tasks including linear, circular and arbitrary trajectories are well explored in computer simulation utilizing MATLAB. These results show that both the mobile base and arms asymptotically track any smooth desired position and velocity trajectories simultaneously, and the coupled dynamics between each subsystem is fully compensated. Four simulation results also illustrate the effectiveness of the proposed models as well as the feasibility of the proposed control scheme. Especially, considering computation loads, transient system responses and tracking errors, the proposed control laws seem to outperform others control strategies, such as neural networks and feedback linearization.
|
|---|
