Analyzing map features and building information for disaster response using agent-based simulations

Abstract

In this paper, we analyze the relationship between map features, including road networks and building information, and the evaluation of multi-agent simulations. In our previous paper, we have defined a movement difficulty for maps that use complexity indexes and have showed that the indexes form a relationship between the results of the multi-agent simulations and road maps[1,2]. In this paper, we define new indexes by considering the building information. We also perform an evaluation experiment to confirm the relationship between the indexes and the results of RoboCup Rescue simulations. The results show that the indexes contribute to the absolute evaluation of agents' cooperative algorithm by considering their environmental dependency. This study in researching agents is absolutely imperative to encourage the development of robotomorphic cooperative agents which act on many kinds of environments. © 2011 Springer-Verlag.