analiza aiqbench
This arXiv preprint proposes a plug-in module that adds calibrated camera geometry to a pretrained vision-language-action robot policy (π0) without changing its action space or training objective. The reported gains on LIBERO, RoboCasa24, and RoboTwin2.0, and the robustness to shifted camera viewpoints, are benchmark and small-scale real-robot results, so their generality still needs independent replication. For readers, the key takeaway is that multi-view consistency can be improved with a modest architectural addition, with supervision from ground-truth point maps or a teacher model rather than depth sensors.
G$^3$VLA: Geometric Inductive Bias for Vision-Language-Action Models
arXiv:2606.24472v2 Announce Type: replace-cross Abstract: 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$^3$VLA, 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$^3$VLA 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 $\pi^3$X teacher predictions, enabling training without depth sensors or manual geometric annotations. On $\pi_0$, G$^3$VLA 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 $\pi_0$ and $\pi_{0.5}$ 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.
Abstract: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$^3$VLA, 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$^3$VLA 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 $\pi^3$X teacher predictions, enabling training without depth sensors or manual geometric annotations. On $\pi_0$, G$^3$VLA 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 $\pi_0$ and $\pi_{0.5}$ 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.
Robotics (cs.RO); Artificial Intelligence (cs.AI)
arXiv:2606.24472 [cs.RO]
(or arXiv:2606.24472v2 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2606.24472 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 313 cuvinte
Subiecte conexe
Language-Augmented Semantic Priors for B-Spline Surface Fitting
arXiv · 12 sept. 2026
Text-Vision Synergistic Token Caching: A Training-Free Framework for Efficient Vision-Language-Action Inference
arXiv · 29 sept. 2026
Multi-Robot Motion Planning from Vision and Language using Heat-Inspired Diffusion
arXiv · 16 iun. 2026
Freeze, Share, Shrink: Rethinking the Action Backbone in Diffusion Policies
arXiv · 16 sept. 2026