Research Of Motion Planning Of Manipulator In High-dimensional Configuration Space Based On ROS

The motion planning of the manipulator is a motion sequence problem in a given obstacle environment,in other words,that is a non-collision path of a robot arm from the starting point to the target point.For the six rotating joints of the UR5 manipulator,it is necessary to describe the rotating joints motion problem of the six-dimensional vector[?1,?2,?3,?4,?5,?6]T in the obstacle environment.Taking into account the traditional path planning algorithm in the high-dimensional configuration space exists in the calculation of the amount of "index disaster" and other issues,this paper applies to high-dimensional configuration space within the rapidly random-exploring tree algorithm RRT,through node sampling,collision detection,it generate needed path to avoid the traditional algorithm of a large number of calculations.In this paper,the forward kinematics modeling of the UR5 manipulator with six degrees of freedom is carried out firstly.Because the inverse kinematics is more complicated,in order to obtain its closed solution,based on Pierper criterion,a new method is proposed to fix the 2nd,3rd,4th joints as the 3-R plane arm in the begining,then the x-y plane of the fifth joint is mapped into the space of the first joint coordinate system to solve the 5th,1st joint angles,and the other joint angles are solved by numerical analysis method.The inverse kinematics is validated by substituting the angle value.In order to be able to move in accordance with the ideal trajectory,this paper carries out the straight line and arc trajectory planning in the Cartesian space.The trajectory planning is done by offline,and the correctness of the inverse solution is verified.Secondly,in this paper,a greedy strategy is applied to the heuristic Bi-RRT algorithm to extend the node toward the target point or the obstacle area,so it has a faster step size relative to the RRT.In this paper,a sampling strategy of Gaussian distribution is proposed to replace the original random sampling characteristics,and apply to the RRT * with the optimal path,to improve the convergence speed of the path.In view of the fact that the path is optimal and convergent,the collision detection algorithm is used to determine the integrity and fastness of the sampling planning path.In this chapter,the collision detection of the robot arm and the obstacle is carried out by the method of AABB aligned axis bounding tree.Because of the existence of the collision-free path has jagged shapes,in order to be able to control manipulator without intermittent walking effect,taking the method of spline smoothing to make it more smooth in path sequence,to ensure that the robot can move smoothly.Finally,in the current international mainstream open source robot operating system(ROS)platform,complete UR5 robot arm virtual motion control system design.First,the URDF(United Robot Description File)format file describes the kinematic chain and inertia characteristics of the robot arm and the end gripper,and depicts the parent-child relationship diagram of each connected link and joint.Secondly,through Moveit software package to complete compilation work of set-up file and launch file.in order to simulate the real prototype of the UR5,this paper under the ROS mechanism designed control and simulation of the software architecture,create joint control file,create Moveit control manager to start the joints,simulate the controller to control the motion process of UR5,analyzes the solution flow of forward and inverse kinematics under ROS.Finally,the RRT *-Gaussian sampling algorithm is loaded by the Moveit motion plug-in OMPL,and establish two different obstacles Scene,complete UR5 non-collision autonomous path search process,complete the pick and place work.Finally,other planning algorithms RRT,RRT *,RRT-Connect and RRT-Gaussian are loaded,and the characteristics of each algorithm are analyzed by box diagram,and the motion state information of each joint of robot arm is derived.