Approximate Planning for Decentralized MDPs with Sparse Interactions

We explore how local interactions can simplify the process of decision-making in multiagent systems. We review decentralized sparse-interaction Markov decision process thatexplicitly distinguishes the situations in which the agentsin the team must coordinate from those in which they canact independently. We situate this class of problems withindifferent multiagent models, such as MMDPs and transitionindependent Dec-MDPs. We contribute new algorithm for efficient planning in this class of problems. We provideempirical comparisons between our algorithms and other existing algorithms for this class of problems.

Authors

• Francisco Melo
• Manuela Veloso J.P.Morgan AI Research

Keywords

• Planning under uncertainty
• cooperative multiagent systems
• sparse interaction