Research On Path Planning For Kiwifruit Picking Manipulator

With the advancement of technology,optimizing the kiwifruit picking process in an intelligent manner has become crucial for accelerating agricultural modernization due to the unique growth pattern and short picking cycle of kiwifruit.This paper proposes the utilization of a six-degree-of-freedom robot as a picking manipulator,connections with the kiwifruit picking end-effector,to enhance the efficiency of kiwifruit picking.The main contributions of this work are as follows:(1)To address the path planning problem in automatic kiwifruit picking,an improved version of the RRT* algorithm called GMM-RRT* is introduced.This algorithm combines the GMM and the artificial potential field strategy.The GMM-RRT* algorithm improves the convergence rate by replacing the uniform sampling method with path-guided GMM sampling,which concentrates the sampling nodes around the goal path.Moreover,the node generation method incorporates a goal-guided bias strategy,utilizing information from the goal node,starting node,and sampling node to determine the direction of random tree growth.The GMM-RRT* algorithm reduces path planning time and minimizes the effective path length,thereby enhancing the efficiency of kiwifruit picking.(2)After obtaining the motion path of the manipulator,a multi-objective trajectory planning solution is proposed using the NSGA-III.This solution aims to achieve faster tracking of the defined path with minimal energy consumption and smoother motion.The trajectory planning problem is transformed into a multi-objective optimization constraint problem,considering the limitations of the mechanical structure that lead to long operation time,significant mechanical shaking,and high energy consumption during kiwifruit picking.The optimization goals include time,energy consumption,and jerk.From the obtained Pareto optimal set,the solution with a relatively short time is selected as the optimal solution and applied to the controller of the picking manipulator.The cubic spline interpolation method based on the NSGA-III algorithm is used in the ROS to construct candidate trajectories and further optimize them.Experimental results demonstrate the effectiveness and feasibility of this approach in reducing picking time and improving picking efficiency.