Accelerating Heuristic Search in Spatial Domains

This paper exploits the spatial representation of state space problem graphs to preprocess and enhance heuristic search engines. It combines classical AI exploration with computational geometry. Our case study considers trajectories in the plane. The application domain is general route planning, independent to the underlying vehicle model. The implemented target system is located on a server and answers client’s shortest path queries with respect to a set of global positioning system traces. A graph is constructed from the traces and is compressed while retaining the original information for unfolding the resulting shortest paths. The search space is pruned by the database entries concisely representing all shortest paths that start with a given edge. The latter algorithm is suited for a server application that has to serve a large number of queries from several different clients. The sudden changes in the topology of graph can affect the pre-computed database entries. This paper dis...

• Type: Conference
• Year: 2003
• Where: PUK
• Authors: Stefan Edelkamp, Shahid Jabbar, Thomas Willhalm
• Keywords: Client’s Shortest Path, Computer Science, Database Entries, PUK 2003, Shortest Paths