AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Scalable and General Whole-Body Control for Cross-Humanoid Locomotion

Yufei Xue, Yunfeng Lin, Wentao Dong, Yang Tang, Jingbo Wang, Jiangmiao Pang, Ming Zhou, Minghuan Liu, Weinan Zhang

ICML 2026 · Reinforcement learning › Visual and representation RL · approach: Dexterous manipulation transfer

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
one generalist policy instead of per-robot specialists, zero-shotXHugWBC · 12 simulated humanoidsreturn: approximately 85% of specialist returnMixed
fine-tune the generalist per robot instead of training specialists from scratchXHugWBC Generalist-FT · 12 simulated humanoidsreturn: approximately 10% higher return than specialistImproved
fine-tune the generalist per robot instead of training from scratchXHugWBC Generalist-FT · 12 simulated humanoidssample efficiencyImproved
use GNN or Transformer policy encoder instead of MLP (ablation)XHugWBC · generalist trainingepisodic return and episode lengthImproved
use physics-consistent morphological randomization instead of naive box-primitive randomization (ablation)XHugWBC · zero-shot unseen robotssurvival rateImproved
use hybrid-mask Transformer with embodiment graph vs MetaMorph policyXHugWBC · zero-shot unseen robotssurvival rateImproved
use MorAL simple architecture and proprioception only vs XHugWBCMorAL · zero-shot unseen robotssurvival rateHurt
fine-tune MetaMorph and MorAL baselinesMetaMorph, MorAL · per-robot fine-tuningreturnMixed
zero-shot deploy sim-trained policy on real robotsXHugWBC · 7 real humanoidsforward velocity tracking error: Dobot Atom 0.063 -> 0.131 m/sHurt
generalist vs specialist velocity trackingXHugWBC · 12 simulated humanoids averageEv_x error: 0.060 -> 0.084Hurt
generalist vs specialist pelvis angle trackingXHugWBC · 12 simulated humanoids averagepelvis angle error: 0.138 -> 0.097Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy