A framework of an agent-based model using social and physical interaction for vulnerability analysis on flood events

Urban hazards have become a matter of concern with both property damages andhuman victims. For flood evacuation, the effective mechanism of residents'communication with other people and sensing methods for the environmental situationswould be essential. In this study, agents' interactions with other agents and the physicalenvironment were discussed for the evacuation simulation of flood hazard using an agentbasedmodel. The knowledge query manipulation language (KQML) was used for asocial interaction that is the communication between agents. The reinforcement learning(RL) was used for a physical interaction that is the agents' perception of the physicalsituation. A physical interaction is an internal interaction since the judgment is based onagent's own experience and memory. A social interaction is an external interaction sincethe judgment is based on the communication with other people. Therefore, this study willprovide a framework of evacuation simulation for urban hazard such as flooding witheffective agent's interaction tools with other agents and the physical environment.