基于TSP_RRT算法的柑橘多目标连续采摘路径规划OA北大核心CSTPCD

[Objective]The harvesting of seasonal fruits such as citrus accounts for about 40%of the total workload,which is time-consuming and labor-intensive.The development of citrus picking robots has become an important way to improve production efficiency.In order to solve the problems of low planning efficiency and long planning path of citrus picking manipulators in the unstructured environment of orchards,an optimal path planning algorithm(TSP_RRT)for continuous citrus picking combining the traveling salesman problem(TSP)and the rapid-exploration random tree(RRT)algorithm is proposed.[Method]In order to describe the real citrus fruit tree in the simulation,a citrus tree model based on geometric envelope method was established.The obstacle portion of the citrus tree,such as the branch trunk,was wrapped in a segmented cylinder,and the citrus that did not need to be picked was wrapped in an encircling sphere.Using the probabilistic random sampling strategy with preconditions,the blindness of sampling can be effectively reduced by introducing the sampling threshold and the generated sampling random number.The target gravity was introduced to control the shortest distance for node growth,thus the growth direction of the expanded tree was always toward the target point.The introduction of adaptive step size allowed the expanded tree to automatically adjust the step size according to the density of obstacles when growing.The target gravity and adaptive step size strategy improved the convergence speed and planning efficiency of the picking robot path planning.In order to improve the overall optimal planning of citrus multi-objective picking path and shorten the path length,the obstacle factor was introduced to obtain the optimal solution of multi-objective picking sequence based on the traveling salesman problem of genetic algorithm,and considering the interference of obstacles such as branches in the picking process.[Result]The simulation results showed that the length of the continuous picking path obtained by the TSP_RRT algorithm planning was shortened by 34.52%and 10.19%respectively compared with the multi-target continuous picking path of the TSP_RRT-connect algorithm and TSP_RRT* algorithm,and the planning time was reduced by about 31%and 50%respectively.The path planning success rate of the TSP_RRT algorithm was approximately 98.8%.[Conclusion]Compared with various RRT algorithms for robotic arm path planning,the improved TSP_RRT algorithm can quickly and accurately generate an optimal path for manipulators picking,shorten the length of path planning and reduce path planning time.This algorithm can provide reference and technical support for citrus picking robots in multi-objective continuous picking.