GesVLA: Gesture-Aware Vision-Language-Action Model Embedded Representations

TL;DR

GesVLA introduces a gesture-aware vision-language-action model that enhances robot manipulation by integrating gesture cues, improving spatial understanding and interaction efficiency in complex scenes.

Contribution

The paper proposes a novel dual-VLM architecture that encodes gesture features into the latent space for better reasoning and action generation, along with a scalable gesture data pipeline and a two-stage training strategy.

Findings

Gesture integration improves target grounding accuracy.

Model enhances human-robot interaction in cluttered environments.

Experimental validation on real-world robotic tasks shows significant performance gains.

Abstract

Vision-Language-Action (VLA) models have shown strong potential for general-purpose robot manipulation by unifying perception and action. However, existing VLA systems primarily rely on textual instructions and struggle to resolve spatial ambiguity in complex scenes with multiple similar objects. To address this limitation, we introduce gesture as a parallel instruction modality and propose a Gesture-aware Vision-Language-Action model (GesVLA). Our approach encodes gesture features directly into the latent space, enabling them to participate in both high-level reasoning and low-level action generation, and adopts a dual-VLM architecture to achieve tight coupling between gesture representations and action policies. At the data level, we construct a scalable gesture data generation pipeline by rendering hand models onto real-world scene images. This reduces the sim-to-real visual gap while producing rich data with diverse motion patterns and corresponding pointing annotations. In addition, we employ a two-stage training strategy to equip the model with both gesture perception and action prediction capabilities. We evaluate our approach on multiple real-world robotic tasks, including a controlled block manipulation task for validation and more practical scenarios such as product and produce selection. Experimental results show that incorporating gesture consistently improves target grounding accuracy and human-robot interaction efficiency, especially in complex and cluttered environments. Project page: https://gwxuan.github.io/GesVLA/.