Abstract: In response to the slow convergence rate, poor flexibility, and the question of the ant colony algorithm getting trapped in local optima, a bidirectional interactive multi-step ant colony algorithm is proposed for research on path planning. Firstly, a dual population approach is adopted, where exploration is conducted through both forward and reverse directions using the ant colony, and an adaptive step-length strategy is implemented to address the issues of local optima trapping and the inflexibility of ants. Secondly, the algorithm is optimized using an adaptive ant population strategy and an improved heuristic function, with the node distance index of the bidirectional ant colony guiding the algorithm’s node transitions to accelerate convergence. Experimental simulation data indicates that the bidirectional interactive multi-step ant colony algorithm not only achieves global fast convergence but also exhibits high stability and shorter computational time. The quality and convergence speed of the obtained solution is superior to those of the improved ant colony algorithm and the multi-step ant colony algorithm based on terminal distance indicators compared in the reference.

Key words: path planning, ant colony algorithm, adaptive step-length, bidirectional interaction, node distance index, heuristic function

摘要： 针对蚁群算法收敛速度较慢、蚂蚁灵活度差以及易陷入局部最优解的问题，提出了一种双向交互多步蚁群算法（bidirectional interactive multi-step ant colony algorithm，BI-MSACO）用于路径规划研究。构建正向与反向双种群使用双向蚁群探索，采用自适应步长策略，解决算法陷入局部最优和蚂蚁灵活度不高的问题。使用自适应蚁群种群数量策略和改进启发式函数对算法进行优化，用双向蚁群的节点距离指数来指导算法节点转移，加快算法收敛速度。经实验仿真数据表明，该研究的双向交互多步蚁群算法在路径规划问题上，不仅可以全局快速收敛，而且具有高度稳定性和更短的运算时间，得到的解的质量和收敛速度相较于参考文献中对比的改进蚁群算法、基于终端距离指标的多步蚁群算法更具优越性。

关键词: 路径规划, 蚁群算法, 自适应步长, 双向交互, 节点距离指数, 启发式函数