Saved in:
| Main Authors: | , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2509.22970 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915537924128768 |
|---|---|
| author | Zhao, Siheng Mao, Jiageng Chow, Wei Shangguan, Zeyu Shi, Tianheng Xue, Rong Zheng, Yuxi Weng, Yijia You, Yang Seita, Daniel Guibas, Leonidas Zakharov, Sergey Guizilini, Vitor Wang, Yue |
| author_facet | Zhao, Siheng Mao, Jiageng Chow, Wei Shangguan, Zeyu Shi, Tianheng Xue, Rong Zheng, Yuxi Weng, Yijia You, Yang Seita, Daniel Guibas, Leonidas Zakharov, Sergey Guizilini, Vitor Wang, Yue |
| contents | We introduce RoLA, a framework that transforms any in-the-wild image into an interactive, physics-enabled robotic environment. Unlike previous methods, RoLA operates directly on a single image without requiring additional hardware or digital assets. Our framework democratizes robotic data generation by producing massive visuomotor robotic demonstrations within minutes from a wide range of image sources, including camera captures, robotic datasets, and Internet images. At its core, our approach combines a novel method for single-view physical scene recovery with an efficient visual blending strategy for photorealistic data collection. We demonstrate RoLA's versatility across applications like scalable robotic data generation and augmentation, robot learning from Internet images, and single-image real-to-sim-to-real systems for manipulators and humanoids. Video results are available at https://sihengz02.github.io/RoLA . |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_22970 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Robot Learning from Any Images Zhao, Siheng Mao, Jiageng Chow, Wei Shangguan, Zeyu Shi, Tianheng Xue, Rong Zheng, Yuxi Weng, Yijia You, Yang Seita, Daniel Guibas, Leonidas Zakharov, Sergey Guizilini, Vitor Wang, Yue Robotics Computer Vision and Pattern Recognition Machine Learning We introduce RoLA, a framework that transforms any in-the-wild image into an interactive, physics-enabled robotic environment. Unlike previous methods, RoLA operates directly on a single image without requiring additional hardware or digital assets. Our framework democratizes robotic data generation by producing massive visuomotor robotic demonstrations within minutes from a wide range of image sources, including camera captures, robotic datasets, and Internet images. At its core, our approach combines a novel method for single-view physical scene recovery with an efficient visual blending strategy for photorealistic data collection. We demonstrate RoLA's versatility across applications like scalable robotic data generation and augmentation, robot learning from Internet images, and single-image real-to-sim-to-real systems for manipulators and humanoids. Video results are available at https://sihengz02.github.io/RoLA . |
| title | Robot Learning from Any Images |
| topic | Robotics Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2509.22970 |