Existing vision-language-action (VLA) models act in 3D real-world but are typically built on 2D encoders, leaving a spatial reasoning gap that limits generalization and adaptability. Recent 3D integration techniques for VLAs either require specialized sensors and transfer poorly across modalities, or inject weak cues that lack geometry and degrade vision-language alignment. In this work, we introduce FALCON (From Spatial to Action), a novel paradigm that injects rich 3D spatial tokens into the action head. FALCON leverages spatial foundation models to deliver strong geometric priors from RGB alone, and includes an Embodied Spatial Model that can optionally fuse depth, or pose for higher fidelity when available, without retraining or architectural changes. To preserve language reasoning, spatial tokens are consumed by a Spatial-Enhanced Action Head rather than being concatenated into the vision-language backbone. These designs enable FALCON to address limitations in spatial representation, modality transferability, and alignment. In comprehensive evaluations across three simulation benchmarks and eleven real-world tasks, our proposed FALCON achieves state-of-the-art performance, consistently surpasses competitive baselines, and remains robust under clutter, spatial-prompt conditioning, and variations in object scale and height.
@article{arxiv.2510.17439,
title = {From Spatial to Actions: Grounding Vision-Language-Action Model in Spatial Foundation Priors},
author = {Zhengshen Zhang and Hao Li and Yalun Dai and Zhengbang Zhu and Lei Zhou and Chenchen Liu and Dong Wang and Francis E. H. Tay and Sijin Chen and Ziwei Liu and Yuxiao Liu and Xinghang Li and Pan Zhou},
journal= {arXiv preprint arXiv:2510.17439},
year = {2026}
}
Comments
Accepted at ICLR 2026. Project page: https://falcon-vla.github.io/