Temporal Representation Alignment: Successor Features Enable Emergent Compositionality in Robot Instruction Following

TL;DR

This paper introduces a method using temporal alignment loss to learn task representations that enable robots to generalize compositionally to new multi-step tasks without explicit planning or reinforcement learning.

Contribution

It demonstrates that temporal representation alignment improves compositional generalization in robotic tasks, even without explicit subtask planning.

Findings

Significant improvement in compositional generalization across robotic tasks.

Effective in both language and goal image task specifications.

Applicable in both real robotic manipulation and simulation environments.

Abstract

Effective task representations should facilitate compositionality, such that after learning a variety of basic tasks, an agent can perform compound tasks consisting of multiple steps simply by composing the representations of the constituent steps together. While this is conceptually simple and appealing, it is not clear how to automatically learn representations that enable this sort of compositionality. We show that learning to associate the representations of current and future states with a temporal alignment loss can improve compositional generalization, even in the absence of any explicit subtask planning or reinforcement learning. We evaluate our approach across diverse robotic manipulation tasks as well as in simulation, showing substantial improvements for tasks specified with either language or goal images.