Learning Object-Centered Autotelic Behaviors with Graph Neural Networks

TL;DR

This paper explores how different graph neural network architectures and goal representations affect the ability of autotelic agents to learn and transfer skills to new, unseen goals, emphasizing object-centered and semantic goal spaces.

Contribution

It introduces a study of various graph neural network policies and goal spaces, demonstrating improved goal-reaching capabilities with object-centered and semantic relational representations.

Findings

Object-centered architectures improve goal transferability.

Semantic relational goals enhance learning of difficult goals.

Graph-based implementations are publicly released for further research.

Abstract

Although humans live in an open-ended world and endlessly face new challenges, they do not have to learn from scratch each time they face the next one. Rather, they have access to a handful of previously learned skills, which they rapidly adapt to new situations. In artificial intelligence, autotelic agents, which are intrinsically motivated to represent and set their own goals, exhibit promising skill adaptation capabilities. However, these capabilities are highly constrained by their policy and goal space representations. In this paper, we propose to investigate the impact of these representations on the learning and transfer capabilities of autotelic agents. We study different implementations of autotelic agents using four types of Graph Neural Networks policy representations and two types of goal spaces, either geometric or predicate-based. By testing agents on unseen goals, we show that combining object-centered architectures that are expressive enough with semantic relational goals helps learning to reach more difficult goals. We also release our graph-based implementations to encourage further research in this direction.