AoE: Always-on Egocentric Human Video Collection for Embodied AI
Bowen Yang ∗1 , Zishuo Li ∗1 , Yang Sun ∗1 , Changtao Miao ∗†1 , Yifan Yang 2 , Man Luo 1 , Xiaotong Yan 1 , Feng Jiang 1 , Jinchuan Shi 3 , Yankai Fu 4 , Ning Chen 4 , Junkai Zhao 5 , Pengwei Wang 5 , Guocai Yao 5 , Shanghang Zhang 4,5 , Hao Chen 3 , Zhe Li 1 , Kai Zhu ‡1 1 Ant Digital Technologies · 2026
Read AoE: Always-on Egocentric Human Video Collection for Embodied AI by Bowen Yang ∗1 , Zishuo Li ∗1 , Yang Sun ∗1 , Changtao Miao ∗†1 , Yifan Yang 2 , Man Luo 1 , Xiaotong Yan 1 , Feng Jiang 1 , Jinchuan Shi 3 , Yankai Fu 4 , Ning Chen 4 , Junkai Zhao 5 , Pengwei Wang 5 , Guocai Yao 5 , Shanghang Zhang 4,5 , Hao Chen 3 , Zhe Li 1 , Kai Zhu ‡1 1 Ant Digital Technologies (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-23893.txt.
Description
Bowen Yang and seventeen co-authors present AoE, an always-on egocentric video collection system for embodied AI, built on people and their phones. The 2026 paper argues that embodied models need vast real-world interaction data, but current collection demands costly hardware and narrow settings, while humans are already ideal embodied agents.
AoE uses an ergonomic neck-mounted phone holder for low-barrier first-person recording, a cross-platform app that processes video on-device, and cloud pipelines that label and filter raw footage into training data. The cloud-edge design lets anyone collect anywhere, at low cost and at scale. Tests on preprocessing quality and downstream tasks show that high-quality egocentric data markedly improve real-world generalization. The system offers a sustainable path past the data scarcity that limits embodied models.
Catalog record
| CREATOR | Bowen Yang ∗1 , Zishuo Li ∗1 , Yang Sun ∗1 , Changtao Miao ∗†1 , Yifan Yang 2 , Man Luo 1 , Xiaotong Yan 1 , Feng Jiang 1 , Jinchuan Shi 3 , Yankai Fu 4 , Ning Chen 4 , Junkai Zhao 5 , Pengwei Wang 5 , Guocai Yao 5 , Shanghang Zhang 4,5 , Hao Chen 3 , Zhe Li 1 , Kai Zhu ‡1 1 Ant Digital Technologies |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 47 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-23893