G^3VLA: Geometric Inductive Bias for Vision-Language-Action Models

Hugging Face Daily Papers Papers
vision-language-action robotics multi-view-geometry camera-calibration manipulation representation-learning

Summary

G³VLA 通过光线嵌入、投影位置编码和跨视角融合，将标定的多相机几何结构注入预训练视觉-语言-动作模型，在不改变动作空间的情况下提升机器人操作的几何一致性与视角鲁棒性。

Vision-language-action (VLA) models have rapidly advanced generalist robot manipulation by harnessing the semantic knowledge of pretrained vision-language backbones. However, their visual representations are typically grounded in 2D image coordinates rather than calibrated camera geometry. This limitation is especially pronounced in multi-camera systems, where views are coupled by known intrinsics and extrinsics but processed as independent images. We propose G^3VLA, a camera-aware geometric module that injects calibrated structure into the visual-token stream of a pretrained VLA without altering its action space or original action-learning objective. G^3VLA combines intrinsic-conditioned ray embeddings, projective positional encoding (PRoPE), and bidirectional cross-view fusion to establish geometrically consistent representations across camera views. Geometric supervision is provided by ground-truth point maps or confidence-gated π^3X teacher predictions, enabling training without depth sensors or manual geometric annotations. On π_0, G^3VLA consistently improves performance on LIBERO and RoboCasa24 under both supervision regimes, while ground-truth supervision improves five-task RoboTwin2.0 averages with three cameras. Experiments on camera-shifted LIBERO-Goal and real robots demonstrate improved robustness to unseen camera viewpoints. Controlled calibration studies across π_0 and π_{0.5} show that these gains depend on correct camera geometry while remaining stable under mild calibration perturbations. These results demonstrate that grounding VLA visual tokens in calibrated camera geometry provides a simple and general mechanism for improving geometric consistency, multi-view reasoning, and viewpoint robustness in pretrained robot policies.
Original Article
View Cached Full Text

Cached at: 10/15/26, 07:04 AM

Paper page - G^3VLA: Geometric Inductive Bias for Vision-Language-Action Models

Source: https://huggingface.co/papers/2606.24472

Abstract

G³VLA injects calibrated multi-camera geometry into vision-language-action models via ray embeddings and cross-view fusion, improving robot manipulation without changing the action space.

Vision-language-action (VLA) models have rapidly advanced generalist robot manipulation by harnessing the semantic knowledge of pretrained vision-language backbones. However, their visual representations are typically grounded in 2D image coordinates rather than calibrated camera geometry. This limitation is especially pronounced in multi-camera systems, where views are coupled by known intrinsics and extrinsics but processed as independent images. We propose G^3VLA, a camera-aware geometric module that injects calibrated structure into the visual-token stream of a pretrained VLA without altering its action space or original action-learning objective. G^3VLA combinesintrinsic-conditioned ray embeddings,projective positional encoding(PRoPE), andbidirectional cross-view fusionto establish geometrically consistent representations across camera views. Geometric supervision is provided by ground-truthpoint mapsor confidence-gated π^3X teacher predictions, enabling training without depth sensors or manual geometric annotations. On π_0, G^3VLA consistently improves performance onLIBEROandRoboCasa24under both supervision regimes, while ground-truth supervision improves five-taskRoboTwin2.0averages with three cameras. Experiments on camera-shiftedLIBERO-Goal and real robots demonstrate improved robustness to unseen camera viewpoints. Controlled calibration studies across π_0 and π_{0.5} show that these gains depend on correct camera geometry while remaining stable under mild calibration perturbations. These results demonstrate that grounding VLA visual tokens in calibrated camera geometry provides a simple and general mechanism for improving geometric consistency, multi-view reasoning, and viewpoint robustness in pretrained robot policies.

View arXiv pageView PDFProject pageAdd to collection

Get this paper in your agent:

hf papers read 2606\.24472

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2606.24472 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2606.24472 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2606.24472 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

HiVLA: A Visual-Grounded-Centric Hierarchical Embodied Manipulation System

Hugging Face Daily Papers

HiVLA introduces a hierarchical vision-language-action framework that decouples semantic planning from motor control using a diffusion transformer action expert for improved robotic manipulation. The system combines a VLM planner for task decomposition and visual grounding with a specialized DiT action expert using cascaded cross-attention, outperforming end-to-end baselines particularly in long-horizon tasks and fine-grained manipulation.

MotionVLA: Vision-Language-Action Model for Humanoid Motion

Hugging Face Daily Papers

Proposes MotionVLA, a vision-language-action model for humanoid motion generation using a dual-stream frequency tokenizer that separately encodes pose and physical dynamics, achieving better diversity and consistency.