Width and Inference Based Planners: SIW, BFS (f), and PROBE

Width and Inference Based Planners: SIW, BFS (f), and PROBE
复制标题

基于宽度和推理的规划器:SIW、BFS (f) 和 PROBE

DOI:
--
复制
发表时间:
2014
期刊:
影响因子:
--
通讯作者:
Hector Geffner
Hector Geffner
中科院分区:
--
文献类型:
--
作者:
N. Lipovetzky;Miquel Ramírez;Christian Muise;Hector Geffner

文献摘要

被引文献

相似文献

我们将规划者 SIW、BFS(f) 和 PROBE 纳入 2014 年国际规划竞赛的敏捷赛道,并为同时运行 SIW 和 BFS(f) 的令人满意的赛道提供随时规划者。 SIW 和 BFS(f) 是利用宽度概念的经典规划器(Lipovetzky 和 ​​Geffner 2012),而 PROBE 是标准的最佳优先搜索规划器,它通过抛出“智能”探针来增强节点的扩展,该探针要么达到目标,要么在低多项式时间内快速终止(Lipovetzky 和 ​​Geffner 2011)。 SIW 的基本构建块是迭代宽度过程 (IW),用于一次实现一个原子目标。 IW 通过执行一系列修剪的广度优先搜索,以问题宽度的时间指数运行。规划器 BFS(f) 将 IW 的新颖性度量与贪婪最佳拳头搜索中的有用动作、地标和删除松弛启发法相结合。在下面的部分中,我们介绍算法的基本概念和实现。
We entered the planners SIW, BFS(f), and PROBE to the agile-track of the 2014 International Planning Competition, and an anytime planner for the satisficing track that runs both SIW and BFS(f). SIW and BFS(f) are classical planners that make use of the notion of width (Lipovetzky and Geffner 2012), while PROBE is a standard bestfirst search planner that augments the expansion of a node by throwing an “intelligent” probe which either reaches the goal or terminates quickly in low polynomial time (Lipovetzky and Geffner 2011). The basic building block of SIW is the Iterative Width Procedure (IW) for achieving one atomic goal at a time. IW runs in time exponential in the problem width by performing a sequence of pruned breadth first searches. The planner BFS(f) integrates a novelty measure from IW with helpful-actions, landmarks and deleterelaxation heuristics in a Greedy Best-Fist search. In the following sections we introduce the basic notions of the algorithms and the implementation.