Multi-Resolution Dijkstra Method Based on Multi-Agent Simulation and its Application to Genetic Algorithm for Classroom Optimization

Abstract

The combinatorial optimization problem of university classroom schedule assignments is formulated using multiagent simulation and genetic algorithms in the evaluation and optimization process. The method we propose consists of global and local multiagent planning. Conventional global planning requires setting subgoals manually, which became a bottleneck in optimization. To solve this problem, a multi-resolution Dijkstra method for selected autonomously, assuming eight classrooms as a real University of Tsukuba building and 250 agents, we confirmed the effectiveness of the proposed multi-resolution Dijkstra’s algorithm as for both global and local route selections, compared to the uniform Dijkstra’s method.