Object-Oriented Dynamics Learning through Multi-Level Abstraction

TL;DR

This paper introduces MAOP, a multi-level abstraction framework for object-oriented dynamics learning from raw visuals, improving generalization, interpretability, and planning efficiency in complex environments.

Contribution

The paper proposes a novel three-level learning architecture with relational reasoning for improved object-based dynamics learning and handling partial observability.

Findings

MAOP outperforms previous methods in sample efficiency and generalization.

Learned models enable effective planning in unseen environments.

MAOP produces interpretable, disentangled representations.

Abstract

Object-based approaches for learning action-conditioned dynamics has demonstrated promise for generalization and interpretability. However, existing approaches suffer from structural limitations and optimization difficulties for common environments with multiple dynamic objects. In this paper, we present a novel self-supervised learning framework, called Multi-level Abstraction Object-oriented Predictor (MAOP), which employs a three-level learning architecture that enables efficient object-based dynamics learning from raw visual observations. We also design a spatial-temporal relational reasoning mechanism for MAOP to support instance-level dynamics learning and handle partial observability. Our results show that MAOP significantly outperforms previous methods in terms of sample efficiency and generalization over novel environments for learning environment models. We also demonstrate that learned dynamics models enable efficient planning in unseen environments, comparable to true environment models. In addition, MAOP learns semantically and visually interpretable disentangled representations.