Abstract Hidden Markov Models for Online Probabilistic Plan Recognition
Hung Bui · 2001
Abstract Markov Policy (AMP) is a model for representing the execution of an abstract plan in noisy and uncertain do- mains. Methods for recognising an abstract policy from a se- quence of noisy observations thus can be used for online plan recognition under uncertainty. In this paper, we extend previ- ous work on policy recognition and consider a general type of abstract policies, including those with non-deterministic ter- minating conditions and factored representations of the state space. We analyse the structure of the stochastic model rep- resenting the execution of the general AMP and provide an efficient hybrid Rao-Blackwellised sampling method for pol- icy recognition that scales well with the number of levels in the plan hierarchy. This illustrates that while the stochastic models for plan execution can be complex, they exhibit spe- cial structures which, if exploited, can lead to efficient plan recognition algorithms.