Saved in:
Bibliographic Details
Main Authors: Lee, Seungjae, Jung, Yoonkyo, Chun, Inkook, Lee, Yao-Chih, Cai, Zikui, Huang, Hongjia, Talreja, Aayush, Dao, Tan Dat, Liang, Yongyuan, Huang, Jia-Bin, Huang, Furong
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2511.21690
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909927056867328
author Lee, Seungjae
Jung, Yoonkyo
Chun, Inkook
Lee, Yao-Chih
Cai, Zikui
Huang, Hongjia
Talreja, Aayush
Dao, Tan Dat
Liang, Yongyuan
Huang, Jia-Bin
Huang, Furong
author_facet Lee, Seungjae
Jung, Yoonkyo
Chun, Inkook
Lee, Yao-Chih
Cai, Zikui
Huang, Hongjia
Talreja, Aayush
Dao, Tan Dat
Liang, Yongyuan
Huang, Jia-Bin
Huang, Furong
contents Learning new robot tasks on new platforms and in new scenes from only a handful of demonstrations remains challenging. While videos of other embodiments - humans and different robots - are abundant, differences in embodiment, camera, and environment hinder their direct use. We address the small-data problem by introducing a unifying, symbolic representation - a compact 3D "trace-space" of scene-level trajectories - that enables learning from cross-embodiment, cross-environment, and cross-task videos. We present TraceGen, a world model that predicts future motion in trace-space rather than pixel space, abstracting away appearance while retaining the geometric structure needed for manipulation. To train TraceGen at scale, we develop TraceForge, a data pipeline that transforms heterogeneous human and robot videos into consistent 3D traces, yielding a corpus of 123K videos and 1.8M observation-trace-language triplets. Pretraining on this corpus produces a transferable 3D motion prior that adapts efficiently: with just five target robot videos, TraceGen attains 80% success across four tasks while offering 50-600x faster inference than state-of-the-art video-based world models. In the more challenging case where only five uncalibrated human demonstration videos captured on a handheld phone are available, it still reaches 67.5% success on a real robot, highlighting TraceGen's ability to adapt across embodiments without relying on object detectors or heavy pixel-space generation.
format Preprint
id arxiv_https___arxiv_org_abs_2511_21690
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle TraceGen: World Modeling in 3D Trace Space Enables Learning from Cross-Embodiment Videos
Lee, Seungjae
Jung, Yoonkyo
Chun, Inkook
Lee, Yao-Chih
Cai, Zikui
Huang, Hongjia
Talreja, Aayush
Dao, Tan Dat
Liang, Yongyuan
Huang, Jia-Bin
Huang, Furong
Robotics
Computer Vision and Pattern Recognition
Machine Learning
Learning new robot tasks on new platforms and in new scenes from only a handful of demonstrations remains challenging. While videos of other embodiments - humans and different robots - are abundant, differences in embodiment, camera, and environment hinder their direct use. We address the small-data problem by introducing a unifying, symbolic representation - a compact 3D "trace-space" of scene-level trajectories - that enables learning from cross-embodiment, cross-environment, and cross-task videos. We present TraceGen, a world model that predicts future motion in trace-space rather than pixel space, abstracting away appearance while retaining the geometric structure needed for manipulation. To train TraceGen at scale, we develop TraceForge, a data pipeline that transforms heterogeneous human and robot videos into consistent 3D traces, yielding a corpus of 123K videos and 1.8M observation-trace-language triplets. Pretraining on this corpus produces a transferable 3D motion prior that adapts efficiently: with just five target robot videos, TraceGen attains 80% success across four tasks while offering 50-600x faster inference than state-of-the-art video-based world models. In the more challenging case where only five uncalibrated human demonstration videos captured on a handheld phone are available, it still reaches 67.5% success on a real robot, highlighting TraceGen's ability to adapt across embodiments without relying on object detectors or heavy pixel-space generation.
title TraceGen: World Modeling in 3D Trace Space Enables Learning from Cross-Embodiment Videos
topic Robotics
Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2511.21690