Action-Model Acquisition from Noisy Plan Traces

There is increasing awareness in the planning com-munity that the burden of specifying complete do-main models is too high, which impedes the appli-cability of planning technology in many real-world domains. Although there have been many learning approaches that help automatically creating domain models, they all assume plan traces (training data) are correct. In this paper, we aim to remove this assumption, allowing plan traces to be with noise. Compared to collecting large amount of correct plan traces, it is much easier to collect noisy plan traces, e.g., we can directly exploit sensors to help collect noisy plan traces. We consider a novel so-lution for this challenge that can learn action mod-els from noisy plan traces. We create a set of ran-dom variables to capture the possible correct plan traces behind the observed noisy ones, and build a graphical model to describe the physics of the do-main. We then learn the parameters of the graphi-cal model and acquire the domain model based on the learnt parameters. In the experiment, we empir-ically show that our approach is effective in several planning domains.