关键词: 自动导引车, U型陷阱, 蚁群优化算法, 栅格地图填充

Abstract: In response to the issues of slow convergence speed,susceptibility to local optima,and deadlocks (such as U-shaped traps) in traditional ant colony algorithms for Automatic Guided Vehicle (AGV) path planning,an improved algorithm for the grid map environment has been proposed. This algorithm optimizes the search process by matching and filling U-shaped traps in the grid map before path optimization. First,a grid model of the AGV path planning environment is established. Then,a non-fillable model and various fillable 3×3 subunit grid models are defined,and U-shaped traps and nodes that do not require exploration are iteratively matched and filled on the grid map using two-dimensional convolution. Subsequently,the ant colony algorithm is applied for path planning,effectively avoiding the problems of slow path convergence and local optima caused by ants getting trapped in U-shaped traps. Finally,simulations are conducted in U-shaped trap grid maps,maze environment grid maps and actual campus environment grid maps. The results show that the algorithm has a maximum filling rate of 83.5 % for passable nodes in the raster map. In the path planning, the pathfinding time is reduced by up to 53.85 % and the path length is reduced by 25.12 %, and in general, the optimization effect of the algorithm in time and path length is obvious, especially when dealing with more complex obstacle environments.

Key words: Automatic Guided Vehicle (AGV), U-shaped trap, Ant Colony Optimization (ACO) algorithm, filled grid map