Language-Driven Representation Learning for Robotics

TL;DR

This paper introduces Voltron, a novel language-driven representation learning framework that leverages human videos and captions to improve visual representations for diverse robotic tasks, outperforming existing methods.

Contribution

The paper presents Voltron, a new framework combining visual reconstruction and language grounding, and provides a comprehensive evaluation suite for robotic visual representations.

Findings

Voltron outperforms prior state-of-the-art methods across five robotic tasks.

Language-driven features improve high-level semantic understanding in robotic perception.

Existing methods show inconsistent results across different robotic vision tasks.

Abstract

Recent work in visual representation learning for robotics demonstrates the viability of learning from large video datasets of humans performing everyday tasks. Leveraging methods such as masked autoencoding and contrastive learning, these representations exhibit strong transfer to policy learning for visuomotor control. But, robot learning encompasses a diverse set of problems beyond control including grasp affordance prediction, language-conditioned imitation learning, and intent scoring for human-robot collaboration, amongst others. First, we demonstrate that existing representations yield inconsistent results across these tasks: masked autoencoding approaches pick up on low-level spatial features at the cost of high-level semantics, while contrastive learning approaches capture the opposite. We then introduce Voltron, a framework for language-driven representation learning from human videos and associated captions. Voltron trades off language-conditioned visual reconstruction to learn low-level visual patterns, and visually-grounded language generation to encode high-level semantics. We also construct a new evaluation suite spanning five distinct robot learning problems $\unicode{x2013}$ a unified platform for holistically evaluating visual representations for robotics. Through comprehensive, controlled experiments across all five problems, we find that Voltron's language-driven representations outperform the prior state-of-the-art, especially on targeted problems requiring higher-level features.