Robust Goal Recognition with Operator-Counting Heuristics

Goal recognition is the problem of inferring the correct Operator-counting heuristics provide a unifying framework goal towards which an agent executes a plan, for a variety of sources of information from planning heuristics given a set of goal hypotheses, a domain model, [Hoffmann that provide both an estimate ofet al., 2004] and a (possibly noisy) sample of the plan being the total cost of a goal from any given state and and indication executed. This is a key problem in both cooperative of the actual operators likely to be in such plans. This and competitive agent interactions and recent information proves to be effective at differentiating between approaches have produced fast and accurate goal goal hypotheses in goal recognition, as we empirically show recognition algorithms.