LTI Agent Rescue: A Partial Global Approach for the RoboCup Rescue Task Allocation

Coordenação é um dos principais problemas em sistemas multiagentes, além de também desempenhar um papel essencial na gestão de desastres. A alocação de tarefas é uma fase importante do problema de coordenação, já que a decomposição de um objetivo em tarefas é a forma mais natural para se organizar o trabalho entre agentes. Nesse artigo, é proposta uma abordagem híbrida de alocação de tarefas para coordenar os agentes na RoboCup Rescue Agent Simulation, que considera a existência de informações locais e globais. Além disso, é apresentada uma metodologia para a análise comparativa entre times de agentes, que é aplicada na comparação dos resultados obtidos pelo time LTI Agent Rescue, que utiliza a abordagem proposta, com os resultados obtidos por outros times de agentes que implementam abordagens distintas para solucionar o problema da coordenação. Abstract: Coordination is one of the key issues in multiagent systems and it also plays an essential role in disaster management. Task allocation is an important phase of the coordination problem, since the decomposition of the objective into tasks is the most natural way to organize work among agents. In this paper, we propose a hybrid task allocation approach to coordinate the agents in the RoboCup Rescue Agent Simulation that considers the existence of both local and global information. Moreover, we provide a methodology to compare agent team results, that is used to compare the results of the LTI Agent Rescue team, which uses the proposed approach, with the results of other rescue agent teams that implement different approaches to solve the coordination problem. Coordination is one of the key issues in multiagent systems and it also plays an essential role in disaster management. Task allocation is an important phase of the coordination problem, since the decomposition of the objective into tasks is the most natural way to organize work among agents. In this paper, we propose a hybrid task allocation approach to coordinate the agents in the RoboCup Rescue Agent Simulation that considers the existence of both local and global information. Moreover, we provide a methodology to compare agent team results, that is used to compare the results of the LTI Agent Rescue team, which uses the proposed approach, with the results of other rescue agent teams that implement different approaches to solve the coordination problem.