Learning STRIPS Action Models with Classical Planning

TL;DR

This paper introduces a flexible method that learns STRIPS action models by translating the problem into classical planning, accommodating various input knowledge levels and partial models.

Contribution

It presents a novel compilation approach that enables learning and validation of STRIPS models from diverse and incomplete input data.

Findings

Effective learning from plans and states

Supports partial and fully specified models

Validates plan execution against models

Abstract

This paper presents a novel approach for learning STRIPS action models from examples that compiles this inductive learning task into a classical planning task. Interestingly, the compilation approach is flexible to different amounts of available input knowledge; the learning examples can range from a set of plans (with their corresponding initial and final states) to just a pair of initial and final states (no intermediate action or state is given). Moreover, the compilation accepts partially specified action models and it can be used to validate whether the observation of a plan execution follows a given STRIPS action model, even if this model is not fully specified.