Anchor Search: A Unified Framework for Unbounded Bidirectional Search

Abstract

In recent years, significant strides in optimal bidirectional heuristic search (Bi-HS) have deepened our theoretical understanding and boosted performance. Yet, algorithms for Bi-HS in unbounded suboptimal scenarios remains largely unexplored. Despite leveraging front-to-end (F2E) and front-to-front (F2F) bidirectional search for optimal algorithms, adapting them for unbounded suboptimal search remains an open challenge. We introduce a novel framework for suboptimal Bi-HS, called anchor search, and use it to derive new algorithms. Additionally, we propose using pattern databases (PDBs) as differential heuristics (DHs) to construct F2F heuristics—a necessity for F2F searches. Our experiments evaluate six anchor search algorithms across diverse domains, with a subset of them outperforming existing methods.