Visual Based Object Grasping And Dynamic Control For Mobile Manipulator

During recent decades, with its rapid development and wide usage in various areas, robotic technology promotes the development of social productive force enormously and improves the people's living standards extensively. Combining abilities of mobile robotics'perception, orientation & navigation, autonomous mobility and flexible operation ability, mobile manipulators are widely used. However, mobile manipulator is a complexly nonlinear system with strong coupling,but also a hot spot in robotics research area presently. Therefore, based on issues of the motion planning and dynamic control of manipulators, this thesis mainly investigates trajectory tracking of manipulators with uncertainties. Moreover, in order to meet a mobile manipulators application, the visual servo control of mobile manipulators is studied.Firstly, a new second order sliding mode control method for robust control of pan-tilt joint of modular manipulator, which makes use of robustness of sliding mode control and hides the switch control law into the integrator to get a continuous control signal, is proposed. Therefore, the high frequency chatter in sliding mode control method can be restrained. Moreover, by choosing integrator initial values, selecting virtual control law, and using saturation function in virtual control law, this thesis improves the method of designing second order sliding mode control. The modified controller reduces chatter problems and steady-state errors. Also, one can design a low frequency controller by using the proposed method, which relaxes the limit of hardware equipment for controllers. Furthermore, simulation results demonstrate the validity and effectiveness of proposed control method.Secondly, to deal with trajectory tracking problems of n-link rigid manipulators with uncertainties, a new adaptive fuzzy dynamic surface control strategy is presented. Initially, an adaptive fuzzy system is utilized to approach uncertainties of the original system. At the same time, an accurate mathematic model of manipulator can be obtained. Then the controller of the obtained model can be designed via the dynamic surface control method. Consequently, this approach can not only ensure the stability of the system, but also get higher accuracy. Moreover, the stability of the system is strictly proved in theory. And simulation results are given to show the strong robustness for uncertainties caused by modeling error, external disturbance, load fluctuation, and the effectiveness of the proposed control strategy.Lastly, aimed at practical application, an integrated planning and control demo of mobile manipulator is raised. In this demo, the "hand", "eye" and "foot" of mobile manipulator work together to complete a task, i.e., the mobile manipulator searches target and then moves there to get it back. And an assignment decomposition strategy is presented in this thesis. The constructed strategy can plan and control the mobile robot and manipulator dispersedly. On the one side, based on the idea of visual servo, images from pan-tilt camera are used as feedback signals to correct the trajectory of mobile robot directly. On the other side, images from wireless camera that local on the end of the manipulator are utilized for computing the accurate position of targets and plan the trajectory of manipulator and execute it ultimately. Furthermore, the established scheme is experimented on the MT-Arm mobile manipulator and excellent performances are obtained.