AirExo-2: Scaling up Generalizable Robotic Imitation Learning with Low-Cost Exoskeletons

Hongjie Fang, Chenxi Wang, Yiming Wang, Jingjing Chen, Shangning Xia, Jun Lv, Zihao He, Xiyan Yi, Yunhan Guo, Xinyu Zhan, Lixin Yang, Weiming Wang, Cewu Lu, Hao-Shu Fang
Proceedings of The 9th Conference on Robot Learning, PMLR 305:198-220, 2025.

Abstract

Scaling up robotic imitation learning for real-world applications requires efficient and scalable demonstration collection methods. While teleoperation is effective, it depends on costly and inflexible robot platforms. In-the-wild demonstrations offer a promising alternative, but existing collection devices have key limitations: handheld setups offer limited observational coverage, and whole-body systems often require fine-tuning with robot data due to domain gaps. To address these challenges, we present AirExo-2, a low-cost exoskeleton system for large-scale in-the-wild data collection, along with visual adaptors that transform collected data into pseudo-robot demonstrations suitable for policy learning. We further introduce RISE-2, a generalizable imitation learning policy that fuses 3D spatial and 2D semantic perception for robust manipulations. Experiments show that RISE-2 outperforms prior state-of-the-art methods on both in-domain and generalization evaluations. Trained solely on adapted in-the-wild data produced by AirExo-2, RISE-2 achieves comparable performance to policies trained with teleoperated data, highlighting the effectiveness and potential of AirExo-2 for scalable and generalizable imitation learning.

Cite this Paper

BibTeX
@InProceedings{pmlr-v305-fang25a, title = {AirExo-2: Scaling up Generalizable Robotic Imitation Learning with Low-Cost Exoskeletons}, author = {Fang, Hongjie and Wang, Chenxi and Wang, Yiming and Chen, Jingjing and Xia, Shangning and Lv, Jun and He, Zihao and Yi, Xiyan and Guo, Yunhan and Zhan, Xinyu and Yang, Lixin and Wang, Weiming and Lu, Cewu and Fang, Hao-Shu}, booktitle = {Proceedings of The 9th Conference on Robot Learning}, pages = {198--220}, year = {2025}, editor = {Lim, Joseph and Song, Shuran and Park, Hae-Won}, volume = {305}, series = {Proceedings of Machine Learning Research}, month = {27--30 Sep}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v305/main/assets/fang25a/fang25a.pdf}, url = {https://proceedings.mlr.press/v305/fang25a.html}, abstract = {Scaling up robotic imitation learning for real-world applications requires efficient and scalable demonstration collection methods. While teleoperation is effective, it depends on costly and inflexible robot platforms. In-the-wild demonstrations offer a promising alternative, but existing collection devices have key limitations: handheld setups offer limited observational coverage, and whole-body systems often require fine-tuning with robot data due to domain gaps. To address these challenges, we present AirExo-2, a low-cost exoskeleton system for large-scale in-the-wild data collection, along with visual adaptors that transform collected data into pseudo-robot demonstrations suitable for policy learning. We further introduce RISE-2, a generalizable imitation learning policy that fuses 3D spatial and 2D semantic perception for robust manipulations. Experiments show that RISE-2 outperforms prior state-of-the-art methods on both in-domain and generalization evaluations. Trained solely on adapted in-the-wild data produced by AirExo-2, RISE-2 achieves comparable performance to policies trained with teleoperated data, highlighting the effectiveness and potential of AirExo-2 for scalable and generalizable imitation learning.} }
Endnote
%0 Conference Paper %T AirExo-2: Scaling up Generalizable Robotic Imitation Learning with Low-Cost Exoskeletons %A Hongjie Fang %A Chenxi Wang %A Yiming Wang %A Jingjing Chen %A Shangning Xia %A Jun Lv %A Zihao He %A Xiyan Yi %A Yunhan Guo %A Xinyu Zhan %A Lixin Yang %A Weiming Wang %A Cewu Lu %A Hao-Shu Fang %B Proceedings of The 9th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2025 %E Joseph Lim %E Shuran Song %E Hae-Won Park %F pmlr-v305-fang25a %I PMLR %P 198--220 %U https://proceedings.mlr.press/v305/fang25a.html %V 305 %X Scaling up robotic imitation learning for real-world applications requires efficient and scalable demonstration collection methods. While teleoperation is effective, it depends on costly and inflexible robot platforms. In-the-wild demonstrations offer a promising alternative, but existing collection devices have key limitations: handheld setups offer limited observational coverage, and whole-body systems often require fine-tuning with robot data due to domain gaps. To address these challenges, we present AirExo-2, a low-cost exoskeleton system for large-scale in-the-wild data collection, along with visual adaptors that transform collected data into pseudo-robot demonstrations suitable for policy learning. We further introduce RISE-2, a generalizable imitation learning policy that fuses 3D spatial and 2D semantic perception for robust manipulations. Experiments show that RISE-2 outperforms prior state-of-the-art methods on both in-domain and generalization evaluations. Trained solely on adapted in-the-wild data produced by AirExo-2, RISE-2 achieves comparable performance to policies trained with teleoperated data, highlighting the effectiveness and potential of AirExo-2 for scalable and generalizable imitation learning.
APA
Fang, H., Wang, C., Wang, Y., Chen, J., Xia, S., Lv, J., He, Z., Yi, X., Guo, Y., Zhan, X., Yang, L., Wang, W., Lu, C. & Fang, H.. (2025). AirExo-2: Scaling up Generalizable Robotic Imitation Learning with Low-Cost Exoskeletons. Proceedings of The 9th Conference on Robot Learning, in Proceedings of Machine Learning Research 305:198-220 Available from https://proceedings.mlr.press/v305/fang25a.html.

Related Material