Adaptive Tracking Control Of Two Manipulators Based On A Virtual ICub Robot

Although intelligence is the main direction of the development of robot technology,the early industrial mechanical manipulators can only accomplish fixed tasks repeatedlyand mechanically which makes their use somewhat limited. In order to improve theflexibility and intelligence of robots, this thesis does some preliminary research with the aidof the iCub robot. This thesis confines the operator into the closed-loop of control systemand then explores a challenging problem that how to make human and the iCub robotadaptable to each other within the control framework of human-robot-interaction strategy.Thus this thesis may shed a light on reducing the requirement of operators who need tolearn or take the initiative to adapt to robots in the traditional robot applications. The maincontributions of this thesis are given as follows:(1) To simulate the manipulator dynamics, the DH parameter method is adopted tomodel the robot arm with the robotics toolbox in MATLAB. And all the parameters comefrom the kinematics and dynamics of the iCub robot.(2) In order to emulate the designed experimental scene, we regard the robot and theoperator as a whole control system and introduce the operator into the closed-loop ofcontrol system through the force/touch sense feedback device in order to realize thehuman-robot-interaction.(3) Based on the robot dynamic equation, this thesis designs the closed-loop controllerrespectively using the fuzzy control algorithm and adaptive control algorithm. This thesisanalyzes the statistic according to the experimental results, then optimizes and adjusts thecontroller parameters until the robot can fulfill the specified tracking task with the guidanceof the operator.The whole experimental simulation has been completed on the computer. Without thenecessity of knowing the parameters of the dynamical system, the robot manipulator fulfillsthe task of handling an object with the guidance of the operator. The designed adaptivecontroller can avoid the computation of the linear regression matrix and inverse operationof parameter matrix which greatly improves the efficiency of algorithm and the robustnessof system. Also this robot control technology based on human-robot-interaction strategycan overcome such a drawback in the early days that robots can only complete specifiedaction sequence in accordance with preset program. In this way the applications of the robots could be widen to some degree and the intelligence of robots could be enhanced.