Planning in nondeterministic domains under partial observability via symbolic model checking
Planning under partial observability is one of the most significant and challenging planning problems. It has been shown to be hard, both theoretically and experimentally. In this paper, we present a novel approach to the problem of planning under partial observability in non-deterministic domains. We propose an algorithm that searches through a (possibly cyclic) and-or graph induced by the domain. The algorithm generates conditional plans that are guaranteed to achieve the goal despite of the uncertainty in the initial condition, the uncertain effects of actions, and the partial observability of the domain. We implement the algorithm by means of BDD-based, symbolic model checking techniques, in order to tackle in practice the exponential blow up of the search space. We show experimentally that our approach is practical by evaluating the planner with a set of problems taken from the literature and comparing it with other state of the art planners for partially observable domains.