Research On Motion Planning Method For Visual Interaction And LFD Of Manipulator

With the development of robotics technology and social demand,service robots have become a research hotspot.There are many difficulties in the research of service robots,such as complex application tasks,unstructured application environment and wide application objects.In this thesis,the main actuator arm of service robot is taken as the object,and its motion planning and visual interaction are studied.The main work is as follows:In this thesis,the development history of robots is reviewed,and the research content and status quo of service robots are introduced.Then,the contents and difficulties of motion planning of the manipulator are explored based on the kinematics of manipulators,the traditional planning algorithm and intelligent planning algorithm are studied,and the representative Rapidly-exploring Random Tree(RRT)algorithm of the latter is simulated and analyzed,the experimental results show that RRT is not suitable for service robots.In order to satisfy the requirement of intelligent planning for service robots,a motion planning method based on Dynamic Motion Primitives(DMPS)is proposed,which enables the manipulator to learn its own training movement,thus completing the generalization of new tasks.Sample selection rules and obstacle avoidance strategies are designed to enable the manipulator to operate intelligently.Taking a three-degree-offreedom space manipulator as a specific object,the feasibility and accuracy of the sample selection and learning from demonstration(LFD)manipulator method are simulated and analyzed in MATLAB,and its effectiveness is verified by physical experiments on the real manipulator.Then,in order to achieve friendly interaction pattern,a visual interactive manipulator control method is proposed,which enables the operator to control the manipulator's motion indirectly and in real time.Physical experiments are designed to verify the method.Then it is integrated into the sample teaching process of LFD system,and a visual interactive LFD system based on DMPS is obtained.Finally,the performance of the system is analyzed and improved through experiments.