Extending Graphplan to Handle Uncertainty & Sensing Actions
Extending Graphplan to Handle Uncertainty & Sensing Actions
复制标题
扩展 Graphplan 来处理不确定性
DOI：
10.1609/aiide.v13i1.12928
复制
发表时间：
1998
期刊：
影响因子：
--
通讯作者：
David E. Smith
中科院分区：
文献类型：
--
作者：
Daniel S. Weld;Corin R. Anderson;David E. Smith
如果一个智能体没有关于世界状态的完整信息，它必须推理世界的其他可能状态，并考虑它的任何行动是否可以减少不确定性。由权变计划者控制的智能体试图在执行之前生成一个鲁棒的计划，该计划考虑并处理所有可能发生的事情。因此，应急计划可以包括收集稍后用于在不同计划分支之间进行选择的信息的感测动作。遗憾的是，以往的应急计划存在语义混乱、不完整、不科学等缺陷。在本文中，我们描述了SGP，一个解决权变规划问题的图计划的派生。SGP将感知未知命题的值的行为与改变其值的行为区分开来。SGP并不支持CNLP和Cassandra所展示的不完整性形式。此外，SGP相对较快。
If an agent does not have complete information about the world-state, it must reason about alternative possible states of the world and consider whether any of its actions can reduce the uncertainty. Agents controlled by a contingent planner seek to generate a robust plan, that accounts for and handles all eventualities, in advance of execution. Thus a contingent plan may include sensing actions which gather information that is later used to select between di(cid:11)erent plan branches. Unfortunately, previous contingent planners su(cid:11)ered defects such as confused semantics, incompleteness, and ine(cid:14)-ciency. In this paper we describe SGP, a de-scendant of Graphplan that solves contingent planning problems. SGP distinguishes between actions that sense the value of an unknown proposition from those that change its value. SGP does not su(cid:11)er from the forms of incompleteness displayed by CNLP and Cassandra. Furthermore, SGP is relatively fast.