Abstract: The path planning of mobile robots not only requires short path distances, but also avoids excessive turning of paths, serious bumps, and poor environmental adaptability. Therefore, this paper proposes improvement heuristics function based on three factors：path length, number of turns, and smoothness of gradient, comprehensively calculating of transition probability. While improving the pheromone update method, it allocates the pheromone amount on each path according to the three-factor comprehensive index, guides ants to approach the path with the best overall performance. And it proposes a non-uniform initial pheromone method to prevent excessive ants into the dead end. It combines improved map modeling barriers to improve path safety. Simulation and experimental results show that the planning path obtained by the improved algorithm has a great improvement in the overall performance of the three factors, and has a good global search capability and convergence. Adjusting the parameters appropriately can also obtain a path with a prominent characteristic. Both the number of iterations and the calculation time perform better.

Key words: Ant Colony Algorithm（ACA）, inspired function, grid path planning, mobile robot, pheromone

摘要： 移动机器人的路径规划不仅要求路径路程短，还要避免路径转弯过多，颠簸程度严重，环境适应性差等问题，为此提出基于路径长度，转弯次数及坡度平滑性三种因素共同影响的改进启发函数，综合计算转移概率；同时改进信息素更新方式，根据三因素综合指标分配各路径上的信息素量，指导蚂蚁向综合性能最好的路径靠近。并提出一种非均匀初始信息素方法，防止过多蚂蚁走入死路。结合改进的地图建模障碍机制，提高路径的安全性。仿真及实验结果表明，改进算法得到的规划路径在三因素综合性能上具有较大提高，且具有较好的全局搜索能力及收敛性，适当调整参数还能得到某一特性表现突出的路径，且迭代次数和计算时间均表现较优。

关键词: 蚁群算法, 启发函数, 路径规划, 移动机器人, 信息素