• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>International Conference on Automated Planning and Scheduling(ICAPS 2006); 2006; >Optimal STRIPS Planning by Maximum Satisfiability and Accumulative Learning
【24h】

Optimal STRIPS Planning by Maximum Satisfiability and Accumulative Learning

机译：通过最大程度的满意度和累积学习来优化STRIPS规划

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Planning as satisfiability (SAT-Plan) is one of the best approaches to optimal planning, which has been shown effective on problems in many different domains. However, the potential of the SAT-Plan strategy has not been fully exploited. Following the SAT-Plan paradigm, in this paper we formulate a STRIPS planning problem as a maximum SAT (max-SAT) and develop a general two-phase algorithm for planning. Our method first represents a planning problem as a combinatorial optimization in the form of a SAT compounded with an objective function to be maximized. It then uses a goal-oriented variable selection to force goal-oriented search and a accumulative learnt strategy to avoid to learn a learnt clause multiple times. We integrate our new methods with SATPLAN04. We evaluate and demonstrate the efficacy of our new formulation and algorithm with SATPLAN04 on many well-known real-world benchmark planning problems. Our experimental results show that our algorithm significantly outperforms SATPLAN04 on most of these problems, sometimes with an order of magnitude of improvement in running time.
机译：作为可满足性的计划（SAT-计划）是最佳计划的最佳方法之一，已证明对许多不同领域的问题都有效。但是，SAT计划战略的潜力尚未得到充分利用。根据SAT-Plan范例，本文将STRIPS规划问题公式化为最大SAT（max-SAT），并开发了通用的两阶段规划算法。我们的方法首先将规划问题表示为SAT形式的组合优化，并结合了要最大化的目标函数。然后，它使用面向目标的变量选择来强制进行面向目标的搜索，并使用累积的学习策略来避免多次学习学习的子句。我们将新方法与SATPLAN04集成在一起。我们评估并证明了采用SATPLAN04的新公式和算法对许多著名的现实世界基准计划问题的有效性。我们的实验结果表明，在大多数这些问题上，我们的算法明显优于SATPLAN04，有时在运行时间上有一定程度的改善。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号