EgoAgent: A Joint Predictive Agent Model in Egocentric Worlds
Lu Chen * Yizhou Wang * Shixiang Tang Qianhong Ma Tong He, Wanli Ouyang Xiaowei Zhou Hujun Bao Sida PengState Key Lab of CAD&CG, Zhejiang University The Chinese University of Hong KongShanghai Jiao Tong University Shanghai Artificial Intelligence Laboratory, ∗ Equal contribution Corresponding author · 2025
Read EgoAgent: A Joint Predictive Agent Model in Egocentric Worlds by Lu Chen * Yizhou Wang * Shixiang Tang Qianhong Ma Tong He, Wanli Ouyang Xiaowei Zhou Hujun Bao Sida PengState Key Lab of CAD&CG, Zhejiang University The Chinese University of Hong KongShanghai Jiao Tong University Shanghai Artificial Intelligence Laboratory, ∗ Equal contribution Corresponding author (2025) 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-2502-05857.txt.
Description
Lu Chen and co-authors build EgoAgent, one model that sees, predicts, and acts from a first-person view, in a paper from two thousand twenty-five. Past work trains split models for each skill, so the skills cannot teach each other. The new model weaves states and actions into one sequence inside a single transformer.
Split branches watch and forecast with a time-skewed design that aids joint learning. Tests covered image sorting, future view forecasting, and three-dimensional body motion forecasting. The joint model beat split rivals across those tasks. Code and trained models are set for public release. Readers in robot vision get a human-like loop of see, guess, and act in one net.
Catalog record
| CREATOR | Lu Chen * Yizhou Wang * Shixiang Tang Qianhong Ma Tong He, Wanli Ouyang Xiaowei Zhou Hujun Bao Sida PengState Key Lab of CAD&CG, Zhejiang University The Chinese University of Hong KongShanghai Jiao Tong University Shanghai Artificial Intelligence Laboratory, ∗ Equal contribution Corresponding author |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 49 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-05857