Learning Efficient Abstract Planning Models that Choose What to Predict

TL;DR

This paper introduces a novel method for learning abstract planning models in robotics that selectively predict relevant changes, improving planning efficiency and generalization across diverse domains.

Contribution

It proposes a new approach to learn operators that choose what to predict, addressing limitations of existing symbolic operator learning in robotics.

Findings

Achieves efficient planning in 10 robotics domains

Generalizes to new initial states, goals, and objects

Outperforms previous methods on BEHAVIOR-100 benchmark

Abstract

An effective approach to solving long-horizon tasks in robotics domains with continuous state and action spaces is bilevel planning, wherein a high-level search over an abstraction of an environment is used to guide low-level decision-making. Recent work has shown how to enable such bilevel planning by learning abstract models in the form of symbolic operators and neural samplers. In this work, we show that existing symbolic operator learning approaches fall short in many robotics domains where a robot's actions tend to cause a large number of irrelevant changes in the abstract state. This is primarily because they attempt to learn operators that exactly predict all observed changes in the abstract state. To overcome this issue, we propose to learn operators that 'choose what to predict' by only modelling changes necessary for abstract planning to achieve specified goals. Experimentally, we show that our approach learns operators that lead to efficient planning across 10 different hybrid robotics domains, including 4 from the challenging BEHAVIOR-100 benchmark, while generalizing to novel initial states, goals, and objects.