aiqbench analysis
InfiNoVA addresses a critical bottleneck in Vision-Language-Action (VLA) models by using 3D Gaussian representations to generate dense, geometrically consistent training views from sparse multi-camera data. This approach significantly improves policy robustness against unseen camera perspectives without requiring architectural changes to the underlying model. The method demonstrates that explicit scene reconstruction can effectively mitigate viewpoint dependency, a key challenge for deploying autonomous robots in real-world environments.
InfiNoVA: Infinite Novel View Augmentation for Viewpoint Invariant Robot Policies
arXiv:2609.27734v1 Announce Type: cross Abstract: Vision-Language-Action (VLA) policies often rely strongly on the camera viewpoints seen during training, causing substantial performance degradation when deployed from unseen perspectives. Collecting demonstrations from sufficiently diverse physical viewpoints is expensive and still provides only sparse coverage of the viewpoint space. We introduce InfiNoVA, a data-augmentation framework that converts synchronized multi-camera demonstrations into a dense distribution of geometrically consistent training views. InfiNoVA reconstructs each manipulation trajectory as a time-varying 3D Gaussian representation and renders novel observations from sampled camera poses while preserving the original state-action correspondence. This explicit scene representation improves frame-level fidelity and temporal consistency while reducing task-critical hallucinations observed in generative novel-view synthesis. Across four real-world manipulation tasks, policies trained with InfiNoVA achieve 5.4x higher average success under unseen randomized viewpoints than both VISTA-based augmentation and the unaugmented policy. InfiNoVA further achieves 1.7x higher success than training directly on all five physical camera views. These results show that dense, geometrically grounded viewpoint augmentation provides a practical route toward camera-robust robot policies without modifying the underlying policy architecture.
Abstract:Vision-Language-Action (VLA) policies often rely strongly on the camera viewpoints seen during training, causing substantial performance degradation when deployed from unseen perspectives. Collecting demonstrations from sufficiently diverse physical viewpoints is expensive and still provides only sparse coverage of the viewpoint space. We introduce InfiNoVA, a data-augmentation framework that converts synchronized multi-camera demonstrations into a dense distribution of geometrically consistent training views. InfiNoVA reconstructs each manipulation trajectory as a time-varying 3D Gaussian representation and renders novel observations from sampled camera poses while preserving the original state-action correspondence. This explicit scene representation improves frame-level fidelity and temporal consistency while reducing task-critical hallucinations observed in generative novel-view synthesis. Across four real-world manipulation tasks, policies trained with InfiNoVA achieve 5.4x higher average success under unseen randomized viewpoints than both VISTA-based augmentation and the unaugmented policy. InfiNoVA further achieves 1.7x higher success than training directly on all five physical camera views. These results show that dense, geometrically grounded viewpoint augmentation provides a practical route toward camera-robust robot policies without modifying the underlying policy architecture.
Robotics (cs.RO); Artificial Intelligence (cs.AI)
arXiv:2609.27734 [cs.RO]
(or arXiv:2609.27734v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2609.27734 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
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.
Full text under the publisher's open licence · 265 words
Related coverage
Less Decoder is More Encoder: Geometric Representation Learning from Novel View Synthesis
arXiv · Oct 6, 2026
PIXIE: A Zero-Shot texture-invariant 6D pose estimation framework for unseen objects with assembly defects
arXiv · Jul 21, 2026
PIXIE: A Zero-Shot texture-invariant 6D pose estimation framework for unseen objects with assembly defects
arXiv · Jul 20, 2026
Systematic Evaluation of Novel View Synthesis for Video Place Recognition
arXiv · Jun 16, 2026