Papers
arxiv:2611.22001

DreamFAR: Real-Time and Long-Horizon Interaction with Diverse Objects

Published on Nov 16
Authors:
,
,
,
,
,
,

Abstract

DreamFAR is a robot-native world model trained from scratch on teleoperation data that predicts long-horizon contact-rich manipulation and supports policy evaluation and mixed-data co-training.

World models have shown great potential in serving as a simulator for robot manipulation. However, existing models are either fit to a single environment or lack ability to simulate long-horizon contact-rich events. We present DreamFAR, a world model whose dynamics model is trained from scratch on a large robot-native teleoperation dataset, with no internet-video pretraining. We present an architecture and learning recipe designed for training on robot-native data. DreamFAR is significantly smaller than the high-performance video generation models, yet shows better understanding of contact and causality, and runs 10 FPS on a commercial GPU. Experimentally, DreamFAR significantly outperforms existing models in predicting the long-horizon future. We further demonstrate various applications where DreamFAR can be used, including policy evaluation and mixed-data co-training. Project page: https://www.yimingdou.com/DreamFAR/

Community

Sign up or log in to comment

Get this paper in your agent:

hf papers read 2611.22001
Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash

Models citing this paper 1

Datasets citing this paper 0

No dataset linking this paper

Cite arxiv.org/abs/2611.22001 in a dataset README.md to link it from this page.

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2611.22001 in a Space README.md to link it from this page.

Collections including this paper 0

No Collection including this paper

Add this paper to a collection to link it from this page.