Conditional Effects in Graphplan
Conditional Effects in Graphplan
复制标题
DOI：
--
复制
发表时间：
1998-07
期刊：
影响因子：
--
通讯作者：
Corin R. Anderson;David E. Smith;Daniel S. Weld
中科院分区：
文献类型：
--
作者：
Corin R. Anderson;David E. Smith;Daniel S. Weld
Graphplan由于其极高的性能而引起了人们的极大兴趣，但该算法无法处理比BIPPS更具表达力的动作表示是一个主要限制。特别是，扩展Graphplan来处理条件效果是一项令人惊讶的微妙工作。在本文中，我们描述了可能的替代方案的空间，然后集中在一个特定的方法，我们称之为因子扩展。因子展开将具有条件效应的操作拆分为多个称为组件的新操作，每个组件对应一个条件效应。由于这些操作组件不是独立的，因此因子化扩展使Graphplan的互斥和反向链接阶段变得复杂。作为补偿，因子化扩展通常产生比更明显的完全扩展到排他的CNOPS操作小得多的域模型。我们目前的实验结果表明，因子膨胀占主导地位的充分扩大大的问题。
Graphplan has attracted considerable interest because of its extremely high performance, but the algorithm's inability to handle action representations more expressive than STRIPS is a major limitation. In particular, extending Graphplan to handle conditional effects is a surprisingly subtle enterprise. In this paper, we describe the space of possible alternatives, and then concentrate on one particular approach we call factored expansion. Factored expansion splits an action with conditional effects into several new actions called components, one for each conditional effect. Because these action components are not independent, factored expansion complicates both the mutual exclusion and backward chaining phases of Graphplan. As compensation, factored expansion often produces dramatically smaller domain models than does the more obvious full-expansion into exclusive STRIPS actions. We present experimental results showing that factored expansion dominates full expansion on large problems.