论文标题
使用最长路径的长度计算计划长度边界
Computing Plan-Length Bounds Using Lengths of Longest Paths
论文作者
论文摘要
我们设计了一种方法，可以精确计算出分解状态空间中最长的简单路径的长度，例如经典规划中遇到的状态空间。尽管此问题的复杂性是NEXP-HARD，但我们表明我们的方法可用于计算计划长度的实际有用的上限。我们表明，计算的上限比以前的边界技术所产生的界限要好得多（在许多情况下，数量级），并且可以用于改善基于SAT的计划。
We devise a method to exactly compute the length of the longest simple path in factored state spaces, like state spaces encountered in classical planning. Although the complexity of this problem is NEXP-Hard, we show that our method can be used to compute practically useful upper-bounds on lengths of plans. We show that the computed upper-bounds are significantly (in many cases, orders of magnitude) better than bounds produced by previous bounding techniques and that they can be used to improve the SAT-based planning.
