
Humanoid Robots
G1
SIMPLE: 221 successes in 270 trials across nine Unitree G1 tasks. BSP used 35.0M total parameters. · The 48.3M-parameter BSP succeeded in 8/10 trials on each of the two physical G1 tasks.
Version scope and roles for these associations await verification.
Body Schema Policy (BSP) is a visuomotor imitation-learning policy for whole-body locomotion and manipulation. Joint and link tokens follow the robot’s kinematic tree and query image features to denoise action chunks. The authors evaluated BSP on Unitree G1 in simulation and on hardware and on a simulated Fetch mobile manipulator.
Paper available
Author order: Original order
Mohamed bin Zayed University of Artificial Intelligence · Source
Publication date unknown · Accessed: 2026-11-22