Obstacle Avoidance Motion Planning Of Serial Manipulator Based On Improved Dual-tree RRT Algorithm

With the "Industry 4.0" and "Made in China 2025",the comprehensive strategy focuses heavily on intelligent manufacturing in 10 strategic sectors and has the aim of securing China's position as a global powerhouse in high-tech industries such as robotics,aviation,and new energy vehicles.The serial robotic manipulator has been widely used in manufacturing,although,the motion planning still relies heavily on manual teaching,causing problems such as large labor cost and low automation.This thesis proposes an improved dual-tree RRT(Rapid-Exploration Random Tree)algorithm combined with Dijkstra's algorithm to achieve fast effect of obstacle avoidance motion planning for serial manipulators,i.e.,Dobot3-DOF and PUMA560 6-DOF manipulators.The research develops from mathematical modeling,kinematics analysis,path planning,collision detection,and obstacle avoidance trajectory planning of manipulators.The main research contents of this thesis are as follows:Firstly,the three-dimensional(3D)mechanical and mathematical models of the Dobot and PUMA560 serial manipulator are developed in the MATLAB Robotics Toolbox through the forward and inverse kinematics of the manipulator and the Denavit-Hartenberg(D-H)parameter.Based on the Monte Carlo point cloud and computer graphics,the respective manipulator workspace models are obtained,and the interference relationship between manipulators and spatial obstacles models is described mathematically.Secondly,in view of the shortcoming of low efficiency of the traditional path planning method,an improvement of dual-tree RRT on obstacle-avoidance path planning is proposed with the strategy of changing extended target points,extreme greed,and the novel technique of collision detection.The paths obtained by the improved extension algorithm are performed for secondary optimization using the Dijkstra algorithm,improving the path smoothness.Simulation comparisons between the improved algorithm and the conventional conducted in2 D and 3D show that the improved has advantages in planning time and length of obstacle avoidance paths.Finally,path planning is carried out for the two manipulators in Cartesian space and jointspace respectively,thus the Cartesian path points and the joint space path points are obtained.In order to get time information of the trajectory,the trajectory planning algorithms based on cubic and quintic B-spline interpolation functions are applied respectively to interpolate trajectory about the joint angles mapping from Cartesian space path points and coming from joint space points,and the curves of angle,angular velocity and angular acceleration of each joint about time are obtained and compared.The simulation tests illustrate that the collision-free planning algorithm proposed in this thesis is feasible and of certain theoretical and practical significance for the motion control of serial manipulators.