基于改进蚁群算法的穴盘苗补苗移栽路径规划方法

Path Planning Method for Hole Tray Seedling Transplanting Based on Improved Ant Colony Algorithm

摘要:

为提高温室番茄穴盘苗补苗移栽的工作效率，对补苗移栽路径进行规划，以减少路径规划长度和运算时间，提高机械手补苗效率和缩短反应时间。提出一种基于改进蚁群算法（Improved ant colony optimization）的机械臂补苗移栽路径规划方法，首先，采用多因素启发函数，在启发函数中加入角度因子，增强路径的全局规划性；其次，为解决传统蚁群算法收敛速度慢的问题，引入了自适应挥发系数和动态权重系数；最后针对补苗路径规划背景下信息素复杂无序的问题，在信息素更新下加入边缘距离因子并设置信息素阈值，目的是减少路径规划时间，加快算法收敛。仿真结果表明，相比于传统优化算法，改进蚁群算法能有效优化补苗移栽路径。在试验条件128孔穴盘下，该模型的路径规划长度相比固定顺序法缩短14.65%，相比蚁群算法缩短6.76%，相比遗传算法缩短3.68%，相比克隆选择算法缩短1.01%。对比可知，改进蚁群算法更有利于补苗移栽路径规划，该模型可作为温室穴盘苗机械化补栽路径规划算法控制基础。

Abstract:

In order to improve the efficiency of greenhouse tomato plug seedling transplanting, the transplanting path was planned to reduce the length and computation time of the path planning, improve the efficiency of mechanical arm transplanting, and shorten the reaction time. A path planning method for robotic arm seedling transplanting was proposed based on improved ant colony optimization (improved ACO) algorithm. Firstly, a multi factor heuristic function was adopted, in which an angle factor was added to enhance the global planning of the path. Secondly, to solve the problem of slow convergence speed in traditional ant colony algorithms, adaptive volatility coefficients and dynamic weight coefficients were introduced. Finally, in order to address the problem of complex and disordered pheromones in the context of seedling path planning, edge distance factors were added and pheromone thresholds were set under pheromone updates, with the aim of reducing path planning time and accelerating algorithm convergence. The simulation results showed that compared with traditional optimization algorithms, the improved ant colony algorithm model can effectively optimize the path of seedling transplantation. Under the experimental conditions of 128 hole tray, the path planning length of this model was shortened by 14.65% compared with that of the fixed sequence method, 6.76% compared with that of the ant colony algorithm model, 3.68% compared with that of the genetic algorithm model, and 1.01% compared with that of the clone selection algorithm model. By comparison, it can be seen that improving the ant colony algorithm model was more beneficial for planning the path of transplanting seedlings. This model can serve as the control basis for the path planning algorithm of mechanized transplanting of greenhouse plug seedlings.