Time-Critical Decision Making in Interactive Dynamic Influence Diagram

Time-critical dynamic decision making is a quite challenging task in many real-world applications. It requires to play a trade-off between solution optimality and computational tractability. It is especially true for multiagent settings under uncertainty. In this paper, we model time-critical dynamic decision problem using the representation of interactive dynamic influence diagram (I-DID). We formalize I-DID by providing time-index to nodes within the model. This results in a model that has the ability to represent space-temporal abstraction. In addition, we propose a new method for selecting the abstract model without arbitrarily compromising solution optimality. We evaluate the performance of our method in two benchmark settings and provide results in support.