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

Abdulaziz, Mohammad, Berger, Dominik

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.

扫码加入交流群

加入微信交流群

微信交流群二维码

扫码加入学术交流群,获取更多资源