Multi-Agent Reinforcement Learning-Based Evacuation Models under Emergency

University Transportation Centers Program2022PDFTech ReportYang, YupengYu, JiahaoKarnati, KavyaLiu, DahaiNamilae, SirishPark, HyoshinNorth Carolina A&T Sate University. Transportation Institute. Center For Advanced Transportation MobilityEmbry-Riddle Aeronautical UniversityUnited States. Department of Transportation. Office of the Assistant Secretary for Research and TechnologyUnited States. Department of Transportation. University Transportation Centers (UTC) ProgramNorth Carolina A&T Sate University. Transportation Institute. Center For Advanced Transportation MobilityUnited StatesBehaviorCrowdsDecision makingEvacuationMachine learningPedestrian movementPedestrian trafficRoute choiceSimulationEmergenciesReinforcement LearningMulti-agent collaborationAirport evacuationCATM-2022-R3-ERAU69A3551747125Final Report, 05/01/2019-12/31/2021Reinforcement learning (RL) has been widely used in intelligent transportation systems, especially under emergent situations, RL can explore the dangerous environment and make optimal decisions to guide the evacuation process for human beings. Within RL algorithms, a Multi-agent Reinforcement Learning (MARL) an artificial intelligence-based model that enables multiple agents to communicate and respond to emergent situations for a more efficient exploration and evacuation process, especially when the uncertainty level is high. This study simulated robot agents to explore and evacuate from a dynamic environment with or without collaboration using the MARL Q-learning algorithm. The multi-agent collaboration method was found to perform better than the single-agent exploration regarding the evacuation time, death counts, and reward both in the static threats and dynamic threats environment. Results were discussed, and future directions were given in the end.143