Entropy-Based Victim Search in RoboCup Rescue Agent Simulation

Suddenly fall of unexpected disasters cause serious damage and casualties to our society. In order to minimize damage, search and rescue tasks should be carried out immediately after the disasters. This paper introduces an entropy based search model to RoboCup rescue agent simulation system which enables our search agents to find victims as early as possible. Key point of this proposed model lies in how to determine sequences of search targets from various candidate locations at each simulation step. Detailed illustration is then given to solve this key point using entropy based method. As for agent coordination, commonly used partition method is adopted to eliminate conflicts between agents. Simulation result validates this model when compared to random search and thus offers certain reference to real disaster emergencies.