基于逻辑方法的开放式规划问题研究  

Research on open world planning based on logic method

作　　者：高洁[1,2] 刘亚松[1] 卓汉魁[3] 李磊[2] 

机构地区：[1]吉林大学珠海学院,广东珠海519041 [2]中山大学信息科学与技术学院,广州510275 [3]中山大学超级计算学院,广州510275

出　　处：《计算机应用研究》2016年第5期1298-1302,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(61309011);高校基本科研业务费青年教师重点培育项目(14lgzd06)

摘　　要：在开放世界中求解智能规划问题往往是比较困难的,这是由于在开放世界中,某些对象可能是未知的,因而在搜索规划解时需要考虑不同的可能性。针对开放世界中的规划问题,通常需要使用传感器观察未知的对象,但是传感器无法保证获取所需的所有信息。利用逻辑推理方法,提出了在目标状态信息不完整情况下的智能规划问题求解算法PQG(planner with query-goal),并通过三个规划领域下的实验验证了该算法的有效性。Intelligent planning in an open world was challenging,since existence of certain objects or fluents was often unknown before planning,then it had to take various scenarios into account before searching plans. A common method to solve planning problems in an open world was to depend on sensors to observe the unknown objects or fluents,but sensors were not capable of correctly capturing all information needed for planning. Based on logic method,this paper proposed a novel planning algorithm PQG to solve intelligent planning problem with incomplete goal states information. And the experiment result shows that algorithm PQG is effective in three planning domains.

关 键 词：智能规划 逻辑 开放世界 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]