This paper tackles spatial perception and manipulation challenges in Vision-Language-Action (VLA) models. To address depth ambiguity from monocular input, we leverage a pre-trained multi-view diffusion model to synthesize latent novel views and propose a Geometry-Guided Gated Transformer (G3T) that aligns multi-view features under 3D geometric guidance while adaptively filtering occlusion noise. To improve action learning efficiency, we introduce Action Manifold Learning (AML), which directly predicts actions on the valid action manifold, bypassing inefficient regression of unstructured targets like noise or velocity. Experiments on LIBERO, RoboTwin 2.0, and real-robot tasks show our method achieves superior success rate and robustness over SOTA baselines. Project page: https://junjxiao.github.io/Multi-view-VLA.github.io/.
@article{arxiv.2605.11832,
title = {Learning Action Manifold with Multi-view Latent Priors for Robotic Manipulation},
author = {Junjin Xiao and Dongyang Li and Yandan Yang and Shuang Zeng and Tong Lin and Xinyuan Chang and Feng Xiong and Mu Xu and Xing Wei and Zhiheng Ma and Qing Zhang and Wei-Shi Zheng},
journal= {arXiv preprint arXiv:2605.11832},
year = {2026}
}