Abstract: Task decomposition and task coordination are interrelated and closely integrated process in multi-agent systems task planning.And/or tree is used to indicate the problem statute and solving process in artificial intelligence，with that a complex multi-stage problem is decomposed into several sub-problems which are easily solved.With regards to the complexity of the tasks and timing constraints，a task planning combined weighted and/or tree and AOE-network is presented.According to timing constraints between tasks，complex tasks are decomposed or transformed layer by layer，a structure of weighted and/or tree is built.Then the weighted and/or tree is converted to an AOE-network，concerted planning coordination is executed based on the earliest time of occurrence.Simulation results demonstrate the feasibility and effectiveness of task planning method.

Key words: weighted and/or tree, AOE-network, multi-agent systems, task planning, task decomposition, task coordination

摘要： 多智能体任务规划中的任务分解、任务协调是相互关联和紧密结合的过程。与或树是人工智能中用于表示问题规约以及求解过程的一种方法，它能把复杂的多阶问题分解成多个易于求解的子问题。针对任务的复杂性和时序约束问题，提出一种结合带权与或树和AOE-网的任务规划方法。根据任务的时序约束，对复杂任务进行逐层分解或变换，建立带权与或树结构;将带权与或树转换为AOE-网，进行基于最早发生时间的任务计划一致协调。仿真结果验证了任务规划方法在多智能体系统中的可行性和有效性。

关键词: 带权与或树, AOE-网, 多智能体, 任务 规划, 任务分解, 任务协调