# The Moving Eye: Enhancing VLA Spatial Generalization via Hybrid Dynamic Data Collection

arXiv: https://arxiv.org/abs/2607.02322 · HTML: https://mlsanity.com/paper/2607.02322/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in 3D and spatial-aware VLA, one of 13 directions in Vision-language-action models.
- Builds on 24 of its 26 references that are in the map.
- First posted on arXiv 2 Jul 2026.

## Placed in

- Vision-language-action models / [3D and spatial-aware VLA](https://mlsanity.com/vla/3d-and-spatial-aware-vla/)

## Builds on (most cited first, 24 of 24)

- [Diffusion Policy: Visuomotor Policy Learning via Action Diffusion](https://mlsanity.com/paper/2303.04137/) (2023)
- [π0.5: a Vision-Language-Action Model with Open-World Generalization](https://mlsanity.com/paper/2504.16054/) (2025)
- [Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware](https://mlsanity.com/paper/2304.13705/) (2023)
- [GR00T N1: An Open Foundation Model for Generalist Humanoid Robots](https://mlsanity.com/paper/2503.14734/) (2025)
- [RVT: Robotic View Transformer for 3D Object Manipulation](https://mlsanity.com/paper/2306.14896/) (2023)
- [Project Aria: A New Tool for Egocentric Multi-Modal AI Research](https://mlsanity.com/paper/2308.13561/) (2023)
- [LIBERO-PRO: Towards Robust and Fair Evaluation of Vision-Language-Action Models Beyond Memorization](https://mlsanity.com/paper/2510.03827/) (2025)
- [Decomposing the Generalization Gap in Imitation Learning for Visual Robotic Manipulation](https://mlsanity.com/paper/2307.03659/) (2023)
- [Multi-View Masked World Models for Visual Robotic Manipulation](https://mlsanity.com/paper/2302.02408/) (2023)
- [Novel Demonstration Generation with Gaussian Splatting Enables Robust One-Shot Manipulation](https://mlsanity.com/paper/2504.13175/) (2025)
- [Vision in Action: Learning Active Perception from Human Demonstrations](https://mlsanity.com/paper/2506.15666/) (2025)
- [View-Invariant Policy Learning via Zero-Shot Novel View Synthesis](https://mlsanity.com/paper/2409.03685/) (2024)
- [SPA: 3D Spatial-Awareness Enables Effective Embodied Representation](https://mlsanity.com/paper/2410.08208/) (2024)
- [GeoAware-VLA: Implicit Geometry Aware Vision-Language-Action Model](https://mlsanity.com/paper/2509.14117/) (2025)
- [ActiveUMI: Robotic Manipulation with Active Perception from Robot-Free Human Demonstrations](https://mlsanity.com/paper/2510.01607/) (2025)
- [Do You Know Where Your Camera Is? View-Invariant Policy Learning with Camera Conditioning](https://mlsanity.com/paper/2510.02268/) (2025)
- [VLA Models Are More Generalizable Than You Think: Revisiting Physical and Spatial Modeling](https://mlsanity.com/paper/2512.02902/) (2025)
- [Adversarial Data Collection: Human-Collaborative Perturbations for Efficient and Robust Robotic Imitation Learning](https://mlsanity.com/paper/2503.11646/) (2025)
- [Visual-Policy Learning through Multi-Camera View to Single-Camera View Knowledge Distillation for Robot Manipulation Tasks](https://mlsanity.com/paper/2303.07026/) (2023)
- [ManiVID-3D: Generalizable View-Invariant Reinforcement Learning for Robotic Manipulation via Disentangled 3D Representations](https://mlsanity.com/paper/2509.11125/) (2025)
- [Invariance Co-training for Robot Visual Generalization](https://mlsanity.com/paper/2512.05230/) (2025)
- [RADAR: Benchmarking Vision-Language-Action Generalization via Real-World Dynamics, Spatial-Physical Intelligence, and Autonomous Evaluation](https://mlsanity.com/paper/2602.10980/) (2026)
- [MOVE: A Simple Motion-Based Data Collection Paradigm for Spatial Generalization in Robotic Manipulation](https://mlsanity.com/paper/2512.04813/) (2025)
- [Viewpoint-Agnostic Manipulation Policies with Strategic Vantage Selection](https://mlsanity.com/paper/2506.12261/) (2025)

## Abstract (arXiv)

Vision-Language-Action (VLA) models have shown remarkable promise in generalized robotic manipulation. However, their spatial generalization remains fragile. We argue that simply increasing the number of viewpoints is insufficient. Models often fall into the trap of Shortcut Learning, latching onto spurious correlations (e.g., fixed relative poses between objects or between the camera and robot base) rather than learning true spatial relationships. In this work, we propose a data-centric solution to enhance VLA spatial generalization. We utilize a dual-arm setup where one arm performs manipulation while the other serves as a mobile environmental camera. We systematically evaluate three data distribution patterns: Fixed, Multi-Fixed, and Moving Views. Our findings reveal that a hybrid strategy, combining continuous camera motion with diverse static viewpoints, yields the best performance by substantially reducing spurious correlations while maintaining training stability. Our experiments demonstrate that this strategy mitigates spurious correlations, enabling VLAs to generalize to unseen camera poses and object configurations where simply adding more static viewpoints fails. Crucially, we reveal that the susceptibility to shortcut learning and the struggle with spatial generalization are universal characteristics shared across diverse architectures. Consequently, all evaluated models (ACT, Diffusion, and VLA models including Pi0 and Gr00t) benefit significantly from our mixed data strategy.