DreamDojo: A Generalist Robot World Model from Large-Scale Human Videos
By: Shenyuan Gao, William Liang, Kaiyuan Zheng, Ayaan Malik, Seonghyeon Ye, Sihyun Yu, Wei-Cheng Tseng, Yuzhu Dong, Kaichun Mo, Chen-Hsuan Lin, Qianli Ma, Seungjun Nah, Loic Magne, Jiannan Xiang, Yuqi Xie, Ruijie Zheng, Dantong Niu, You Liang Tan, K.R. Zentner, George Kurian
Published: 2026-02-09
View on arXiv →#cs.AI
Abstract
DreamDojo introduces a generalist robot world model learned from large-scale human videos, enabling efficient reinforcement learning of robotic policies. This framework co-evolves a video world model and a VLA policy, significantly advancing the ability of robots to understand and interact with diverse environments, paving the way for more adaptable and versatile robotic applications.