Discovering Planning Invariants as Anomalies in State Descriptions

Planning invariants are formulae that are true in every reachable state of a planning world. We describe a novel approach to the problem of discovering such invariants—by analyzing only a reachable state of the planning domain, and not its operators. Our system works by exploiting perceived patterns and anomalies in the state description: It hypothesizes that patterns that are very unlikely to have arisen by chance represent features of the planning world. We demonstrate that the number and types of laws we discover are comparable to those discovered by a system that uses complete operator descriptions in addition to a state description.