Hierarchical path computation approach for large graphs

摘要

Time is a critical factor in several path planning problems such as flood emergency rescue operations, escape planning from fires and chemical warfare agents dispersed in large buildings, evacuation from urban areas during natural disasters such as earthquakes, and military personnel movement. We propose a hierarchical path planning algorithm (HIPLA) for real time path planning problems where the computational time is of critical significance. The main idea of HIPLA is to significantly reduce the search space for path computation by searching in a high-level abstraction graph, whose nodes are associated with precomputed risk estimates. The cumulative risk associated with all nodes along a path determines the quality of a path. We present a detailed experimental analysis of HIPLA by comparing it with two well-known approaches viz., shortest path algorithm (SPAH) [1] and Dijkstra''s algorithm with pruning [2] for large node-weighted graphs.