Research On Path Planning Of Master-Slave Apple Picking Manipulator

With the continuous development of information technology,the research and application of intelligent agricultural equipment has been popularized and deepened to cope with the increasing labor costs.However,single-arm harvesting is difficult to meet high efficiency requirements and cannot complete complex tasks,such as collaborative harvesting and heavy lifting.Based on this,aiming at the problems of low efficiency of path planning and collision between two arms in cooperative motion planning of dual manipulators,this paper studies the motion planning of master-slave apple picking manipulator.Firstly,an improved Rapidexploring Random Tree(RRT)path planning algorithm is proposed in this paper.Secondly,the kinematics characteristics and working task mode of the dual manipulator are analyzed,and the hierarchical collision detection model is established.Finally,simulation and experimental research are carried out.The main contents of this paper include:(1)Research on path planning based on RRT algorithm.Aiming at the problems of blind search,difficult convergence and tortuous path generation in traditional RRT algorithm,an improved algorithm including adaptive guidance strategy,sampling adjustment strategy and path correction method is proposed.Firstly,the guidance method is adaptively selected and the degree of target bias is adjusted through previous search experience to avoid falling into local no solution.Secondly,the sampling adjustment strategy is used to improve the sampling success rate in the local and global complex environment.Finally,the loopback and semiloopback problems are solved by adjusting the connection order of path points,and the smooth path is fitted by Bessel curve.The simulation results show that the average time used by the improved algorithm in different scene maps is 87.05 %,79.09 % and 72.46 % of the targetbiased RRT algorithm,RRTgoalbias algorithm and traditional RRT algorithm,and the average search times are 51.05 %,48.79 % and 38.50 % of the three algorithms,which verifies the effectiveness of the improved algorithm in different obstacle scenarios.(2)Kinematic characteristics and task constraints analysis of dual manipulator.Aiming at the problem of task space overlap and cooperation of dual manipulators,the standard D-H parameter table is established and the forward and inverse kinematics equations are derived.The reachable workspace of the manipulator is calculated by Monte Carlo method.Secondly,based on the joint space analysis of arc and straight line trajectory planning methods and applicable scenarios.Then,the working mode of the dual manipulator is divided into loose coordination mode and tight coordination mode.The loose coordination mode is mainly from the picking mode of the manipulator,and the decoupling planning is adopted to analyze the apple selection,picking process and obstacle avoidance method.The tight coordination working mode is collaborative operation,and the constraint relationship of the trajectory of the dual-arm end-effector is analyzed.Finally,the correctness of kinematics equation,trajectory planning method and attitude constraint solution of dual-arm cooperative handling end effector is verified on the simulation platform.(3)Construction of hierarchical bounding box collision detection model.Aiming at the problems of low efficiency,large amount of calculation and redundant calculation of single collision detection algorithm,a hierarchical bounding box collision detection method is proposed.This method combines the axial bounding box method and the directional cylinder method and makes hierarchical calls.That is,the axial bounding box method is used for collision detection first,and if there is a potential collision possibility,the directional cylinder method is used for collision detection.The experimental results show that the correct collision detection rate of the hierarchical bounding box method is 100 %,which is consistent with the directional cylinder detection method and higher than 60.66 % of the axial bounding box method.The detection time is 58.91 % and 127.90 % of the directional cylinder method and the axial bounding box method.This method can reduce redundant calculation and effectively improve the efficiency of collision detection.(4)Apple picking application test.In order to verify the effectiveness of dual-arm picking,a dual-arm picking platform is first built,and the programming,calibration method and picking process of the manipulator are analyzed.Secondly,collision-free and collision-free picking experiments are carried out.Finally,the experimental data are obtained by encoder and the error analysis is carried out.The experimental results show that the arms can run according to the preset trajectory,which is consistent with the simulation results.