Research On Motion Planning Of Multi-degrees Of Freedom Robot Manipulator Based On Study Of Human Control Strategy

Nowadays a significant feature of the development of robots is the increasing intelligence.Under the wave of artificial intelligence,the robot industry has very broad application prospects.At present,most of the manipulators are engaged in single and repeated work in an industrial structured environment,and the task generalization ability is relatively poor when the work scene changes.We think about combining human operation skills and strategies into robot applications to make robot motion planning more efficient and intelligent.In the view of the above background,this paper mainly discusses the motion planning of the manipulator,and takes the Kuka LBR iiwa 7 R800 collaborative manipulator as the research object to study its effective obstacle avoidance in space.The main research contents are as follows:1.In this article,we first carry out the kinematics modeling of the experimental manipulator.Then we analyze forward and inverse kinematics and use the analytical method to solve the inverse kinematics,then we apply the D-H parameter method to correlate the transformation of Cartesian space and joint space,and lead to the concepts of working space and configuration space in motion planning,which provides theoretical support for the following sampling planning space.2.The concept of GMM-C space is proposed.We first introduce the meaning of learning human control strategy,which is the same as learning from demonstration,and analyze the advantages and research status of the method.For the data collected by manual drag teaching,a Gaussian mixture model is used for learning and modeling,with the purpose of fitting a feasible sampling space of 7 joint angles,thereby limiting the sampling space of the subsequent planning tasks to the range of the learned probability model.In this way,learning of human skills is achieved.3.After the modeling of the demonstration data is completed,we adopt a samplingbased motion planning method.The MATLAB simulation results focus on comparing the advantages and disadvantages of the Rapidly-exploring Random Trees algorithm RRT and two improved algorithms RRT-Connect and RRT * based on RRT.Finally,we use the RRT* algorithm with progressive optimal characteristics to sample in the feasible C space,and verify the feasibility of the proposed GMM-C RRT * algorithm on the motion planning simulation platform of the Robot Operating System.The result proves that the manipulator can effectively complete obstacle avoidance tasks in the manner of human skills and improve planning efficiency.