Output details

Exploiting model equivalences for solving interactive dynamic influence diagrams

<22> This paper provides details of a set of exact and approximate algorithms for solving interactive dynamic influence diagrams (I-DID) - a general framework for multiagent planning. The algorithms were developed based on the principle of model equivalence that considers both behavioural and actional equivalence of agents. They represent state-of-the-art I-DID solutions. A notable contribution of this paper is the introduction of the first scalable benchmark to test various algorithms for solving sequential multiagent decision making problems. It is a testbed for the multiple Unmanned Aerial Vehicles (UAV) planning problem, which has been widely adopted in the intelligent agent planning community.