ARMS: Action-Relation Modelling System for Learning Action Models

We present a system for automatically discovering action models from a set of successful observed plans. AI planning requires the definition of an action model using a language such as PDDL as input. However, building an action model from scratch is a difficult and timeconsuming task even for experts. Unlike the previous work in action-model learning, ARMS does not assume complete knowledge of states in the middle of the observed plans; in fact, our approach would work when no or partial intermediate states are given. These example plans are obtained by an observation agent who does not know the logical encoding of actions and the full state information between actions. In a real world application, the cost is prohibitively high in labelling the training examples by manually annotating every state in a plan example from snapshots of an environment. To learn action models, ARMS gathers knowledge on the statistical distribution of frequent sets of actions in the example plans. It then builds a weighted propositional satisfiability (Weighted SAT) problem and solves it using a weighted MAX-SAT solver. We lay the theoretical foundations of the learning problem and evaluate the effectiveness of ARMS empirically.