Abstract: An online path planning method is proposed.First the rule for subgoal mapping is modified；then the diagonal distance to A* search algorithm for local path finding to better adapt the real time requirement is introduced，also a tabu search and diversification strategy is included to overcome the limitation of oscillation and dead circulation，which is always encountered in traditional rolling method.The flexibility and effectiveness are illustrated in the simulation tests.

Key words: rolling window, subgoal mapping, A* search algorithm, diagonal distance, tabu search, diversification

摘要： 未知环境下机器人路径规划要求实时有效，针对传统动态窗口滚动优化算法易出现局部最小、震荡等问题，重新设计了子目标映射法则，并对局部路径规划算法引入对角线距离作为代价函数中的启发式部分，最后引入了禁忌搜索和多样化随机路径选择模块，克服了局部最小和震荡等问题，仿真实验证明了该算法可靠有效。

关键词: 滚动窗口, 子目标映射, A*搜索算法, 对角线距离, 禁忌搜索, 多样化选择