I-BFM

Model family: I-BFM is a behavioral foundation model for humanoid–object interaction. Unsupervised reinforcement learning captures coupled robot, object and contact dynamics. One reward-conditioned policy handles carrying, pushing and kicking, with experiments on a physical Unitree G1.

Papers & reports (1)

I-BFM: Reward-Conditioned Robust Humanoid Interaction via Unsupervised Reinforcement Learning

Paper available

arXiv: 2610.06129
Authors & affiliations (11)

Author order: Original order

  1. Ziqi Han

    Not provided

  2. Yitang Li

    Not provided

  3. Junhan Sun

    Not provided

  4. Fanrong Dong

    Not provided

  5. Yaojie Shen

    Not provided

  6. Lei Ye

    Not provided

  7. Zetong Jing

    Not provided

  8. Yongqi Zhang

    Not provided

  9. Yiming Zhang

    Not provided

First published
2026-10-05
Latest paper revision
Not provided
Venue
arXiv

Covered model versions

Paper ↗PDF ↗

How to access and use

Public code resources
Public access
Model files
Public access

Training / fine-tuning code

Access
Public access

Verified on · 2026-11-21

Code license
CC BY-NC 4.0

Model files

Model files
Unitree G1 · 166.560M stepsAccess
Release information

SourceVerified on · 2026-11-21

Checks before deployment

Available weights or code do not establish plug-and-play robot compatibility. Check the required configuration, data and interfaces in the source documentation.

Deployment and usage resources

Control and inputs / outputs

Model type
Motion & whole-body control

Scope and training

Output

Actions

Robot form

Humanoid

Training

Reinforcement learning

Submit a source or correction