GazeVLA: Learning Human Intention for Robotic Manipulation

TL;DR

GazeVLA introduces a novel approach that models human intention through gaze to improve robotic manipulation, leveraging large-scale human data and chain-of-thought reasoning for better generalization and performance.

Contribution

The paper presents a framework that explicitly learns and transfers human intention via gaze, bridging the embodiment gap and enhancing robotic manipulation capabilities.

Findings

Outperforms strong baselines in simulation and real-world tasks.

Generalizes better across long-horizon and fine-grained tasks.

Achieves state-of-the-art performance in various benchmarks.

Abstract

Embodied foundation models have achieved significant breakthroughs in robotic manipulation, yet they still depend heavily on large-scale robot demonstrations. Although recent works have explored leveraging human data to alleviate this dependency, effectively extracting transferable knowledge remains a significant challenge due to the inherent embodiment gap between human and robot. We argue that the intention underlying human actions can serve as a powerful intermediate representation for bridging this gap. In this paper, we introduce a novel framework that explicitly learns and transfers human intention to facilitate robotic manipulation. Specifically, we model intention through gaze, as it naturally precedes physical actions and serves as an observable proxy for human intent. Our model is first pretrained on a large-scale egocentric human dataset to capture human intention and its synergy with action, followed by finetuning on a small set of robot and human data. During inference, the model adopts a Chain-of-Thought reasoning paradigm, sequentially predicting intention before executing the action. Extensive evaluations in simulation and real-world settings, across long-horizon and fine-grained tasks, and under few-shot and robustness benchmarks, show that our method consistently outperforms strong baselines, generalizes better, and achieves state-of-the-art performance. Project page: https://gazevla.github.io .