【摘要】 条件效果是智能规划处理更具解释性动作描述语言中最难解决的一种类型 .如何扩展当前已有的规划算法 ,使之具有处理包含条件效果的动作描述语言的能力成为了智能规划领域中的研究热点之一 .文章针对一个高效规划器FFv2 .3在条件效果处理中存在的不足做了一些改进 ,提出了一个新的条件效果处理方法CEFF .主要的改进措施有以下两点 :( 1)引入因子扩展法的思想将动作划分为组件 ,以提高对条件效果的处理效率 ;( 2 )在进行启发式估值的图扩展过程中增加对两元互斥关系的判断 ,以避免大部分dead end状态 .因此 ,CEFF的实用性较FFv2 .3要更广泛一些 .更多还原

【Abstract】 Conditional Effects is the most difficult one in dealing with more expressive action description language in AI planning. It becomes one of the AI planning research’s hotspots, that how to expand the existing planning algorithm to handle the action describe language with conditional effects. This paper improves FF v2.3, which is one of the most efficient planners, in handling conditional effects, and brings forward a new approach of handling conditional effects, named CEFF (Conditional Effects FF). Two main improvements of CEFF are as following: (1) adopting the soul of factor expansion, CEFF increases the efficiency of handling CE. (2) in the expanding planning graph phase, CEFF introduces the judgment of exclusive relation, and avoids the great mass of dead-end states. Therefore, CEFF is more practical than FF v2.3.更多还原