| Mobile manipulators have played a more and more important role in the civilian and military use in modern society, such as industry and human living area, and received tremendous attention. The hot spot of their research mainly concentrated on the motion planning and control of mobile manipulators. Since it is difficult to build the mathematical models for multi-link modular mobile manipulators because of the structure complexity, the control strategy derived from the widely researched two-link plane mobile manipulators is no longer valid. It is inevitable that numerous uncertainties exist because of the structure complexity. Therefore, the model-based centralized controller design method doesn't work. Based on the structure characters of modular mobile manipulators, and inquired into the current research situation, the hierarchical structure with decentralized control method is proposed in this dissertation. The planning and control layer is separated into the motion layer and dynamic control layer, the planning and control strategy of'coordinate planning of the kinematics layer, decentralized control of the dynamics layer' is taken. The dynamic control for the mobile robot subsystem and the manipulator subsystem are investigated with the idea of'coordinated planning, divide and rule', and the corresponding control strategy is brought forward.The research work of this dissertation can be summarized as follows:(1) The hierarchical structure with decentralized control of modular mobile manipulators will be brought forward. For the decentralized structure of modular mobile manipulators, it is almost impossible to design the centralized controller based on mathematical model, and the controller structure can be restricted by the plant structure character. Then, based on the fully comparison of kinematics and dynamics of the two-link manipulators with multi-link modular manipulators, the control system performance requirement of modular mobile manipulators will be put forward, and the hierarchical structure with decentralized control will be designed. The planning and control layer of the control system is divided into two layers including kinematics and dynamics. Motion planning can be settled from kinematics aspect, whereas the decentralized control is taken in the dynamic level. The mobile robot and robot manipulator are viewed as two subsystems of the mobile manipulator, and each joint of the modular manipulator can be linear decoupled from each other. Adaptability analysis of the control system structure is given to show the feasibility of the proposed strategy.(2) Based on the designed control system structure, the motion planning problems of redundant mobile manipulators will be investigated. Analytic method for the inverse kinematics solution of mobile manipulators will be proposed. The forward kinematics of the mobile manipulator is deeply analyzed. In face of the certain task of snatching at small balls, an analytic method of inverse kinematics solving will be brought forward. For the manipulator subsystem doesn't satisfy Pieper theory, and some joint will be restricted by certain condition, the task planning problem will be settled.(3) Considering the dynamic control problem, firstly, the trajectory tracking control of mobile robot subsystem will be investigated. Mobile robot belongs to nonholonomic system, and it is relative easy to build mathematical model for this system, whereas there must be some uncertainties such as the modeling errors and the external noise. Based on the approximation ability of RBF neural networks for nonlinear functions, we use the sliding mode control to design the robust adaptive controller. The nominal model of mobile robot is used to design the equivalent controller, and the robust switch item is used to tackle the uncertainty of the system. RBF neural network can be used to adaptively approximate the gains of the robust switch item, and the adaptive tuning laws can be derived via Lyapunov stability theorem.(4) For the dynamic control of robot manipulator subsystem, we first consider the simple two-link planar electrically-driven robot manipulator. Taking the motor character of each joint into consideration, the robot manipulator can be viewed as the cascade system connected by the joint subsystem and the motor subsystem. Based on the measured joint displacement, the joint velocity can be estimated via Luenberger observer. The uncertainty of the motor dynamics is considered, and the controller will be designed by backstepping design method and sliding mode control. This work shows the controller design methodology for simple plant based on mathematical model.(5) Trajectory tracking controller based on ESO for a class of SISO system will be proposed with application to joint space decentralized robust control for PowerCube modular manipulators. With the model reduction of sliding mode control, any nth-order higher system can be reduced into a generalized first-order system with its argument of sliding mode. The modeling uncertainty and the external disturbances are lumped into a'total uncertainty' which can be estimated and compensated in real time via second-order ESO, and the robust controller will be designed by sliding mode control method. The multi-link modular manipulator is decoupled. Based on the proposed control method, joint space controller design for PowerCube modular manipulator will be designed.(6) The discrete decentralized robust controller for PowerCube joint space will be designed based on the linear decoupled model of the joints and the ADRC theory, considering the reference input is affected by noise. In (4), there is a deficiency that the Luenberger observer is just applicable to accurate model of robot manipulator. In this part, we consider the joint space trajectory tracking problem in the situation that both the uncertainties appeared in joint subsystem and motor subsystem. The controller is designed based on backstepping design strategy. The validation of the proposed control strategy is carried out in the pitching direction of a 2-DOF modular manipulator.
|
PDF Full Text Request