Abstract: A method of dynamic window of candidate solution is proposed in this paper for path planning using ant colony algorithm. The method divides the window in which candidate solutions are distributed firmly and uniformly into several blocks, and then lets candidate solutions simulate the special ants to choose a certain block with a certain probability according to the heuristic and the sum of pheromone of the upper connection of all candidate solutions after several iteration of ant colony algorithm. Finally, candidate solutions and pheromone in every blocks are reallocated, so that solution can be refined. Ultimately, ant colony can search a best solution of path. Simulation experiments and comparison results show that the performance of dynamic window of candidate solution is better than the fixed window of candidate solution.

Key words: dynamic window of candidate solution, ant colony algorithm, path planning

摘要： 提出了蚁群路径规划算法中一种动态候选解窗口的方法。该方法首先在固定均匀分布候选解的窗口上划分成若干分块，接着将负责路径规划的主蚁群的候选决策点看作一系列具有候选解属性的蚂蚁，再用该蚁群算法根据每分块上所有候选解上游连接边上的信息素及其启发信息以一定概率动态分布候选决策点，从而细化候选解，最终使蚁群能搜索到一条更好的路径解。仿真实验和对比的结果表明，动态候选解窗口方法比固定均匀分布候选解的方法可获得更优的性能。

关键词: 动态候选解窗口, 蚁群算法, 路径规划