Robot Learning from Any Images

Published in CoRL, 2025

Siheng Zhao, Jiageng Mao, Wei Chow, Zeyu Shangguan, Tianheng Shi, Rong Xue, Yuxi Zheng, Yijia Weng, Yang You, Daniel Seita, Leonidas Guibas, Sergey Zakharov, Vitor Guizilini, Yue Wang

arXiv / PDF / code / project page / BibTeX

Abstract

RoLA is a framework that transforms any in-the-wild image into an interactive, physics-enabled robotic environment. It operates directly on a single image without requiring additional hardware or digital assets. RoLA democratizes robotic data generation by producing massive visuomotor robotic demonstrations within minutes from a wide range of image sources.

BibTeX

@inproceedings{zhao2025robot,
  title     = {Robot Learning from Any Images},
  author    = {Siheng Zhao and Jiageng Mao and Wei Chow and Zeyu Shangguan and Tianheng Shi and Rong Xue and Yuxi Zheng and Yijia Weng and Yang You and Daniel Seita and Leonidas Guibas and Sergey Zakharov and Vitor Guizilini and Yue Wang},
  booktitle = {Conference on Robot Learning (CoRL)},
  year      = {2025},
  eprint    = {2509.22970},
  archivePrefix = {arXiv}
}

Recommended citation: Zhao, S., Mao, J., Chow, W., Shangguan, Z., Shi, T., Xue, R., Zheng, Y., Weng, Y., You, Y., Seita, D., Guibas, L., Zakharov, S., Guizilini, V., & Wang, Y. (2025). Robot Learning from Any Images. CoRL 2025.