Heuristic search for identical payoff Bayesian games

Abstract

Bayesian games can be used to model single-shot decision problems in which agents only possess incomplete information about other agents, and hence are important for multiagent coordination under uncertainty. Moreover they can be used to represent different stages of sequential multiagent decision problems, such as POSGs and DEC-POMDPs, and appear as an operation in many methods for multiagent planning under uncertainty. In this paper we are interested in coordinating teams of cooperative agents. While many such problems can be formulated as Bayesian games with identical payoffs, little work has been done to improve solution methods. To help address this situation, we provide a branch and bound algorithm that optimally solves identical payoff Bayesian games. Our results show a marked improvement over previous methods, obtaining speedups of up to 3 orders of magnitude for synthetic random games, and reaching 10 orders of magnitude speedups for games in a DEC-POMDP context. This not only allows Bayesian games to be solved more efficiently, but can also improve multiagent planning techniques such as top-down and bottom-up algorithms for decentralized POMDPs.