Research map / Vision-language and multimodal / Robot action and navigation
From Language to Locomotion: Retargeting-free Humanoid Control via Motion Latent Guidance
Zhe Li, Yangyang Wei, Boan Zhu, Yibo Peng, Tao Huang, Pengwei Wang, Zhongyuan Wang, Cheng Chi, Chang Xu, Shanghang Zhang
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 changed | Where | Measured | Effect |
|---|---|---|---|
| use latent-driven retargeting-free pipeline instead of decode+retarget (ablation) | RoboGhost · HumanML/Kungfu tracking, IsaacGym | success rate: 0.93 -> 0.97 | Improved |
| use latent-driven pipeline instead of decode+retarget (ablation) | RoboGhost · deployment time | time cost (s): 17.85 -> 5.84 | Improved |
| use diffusion student policy instead of MLP policy (ablation) | RoboGhost · HumanML subset, IsaacGym | success rate: 0.96 -> 0.97 | Improved |
| use diffusion student policy instead of MLP policy (ablation) | RoboGhost · Unseen MotionMillion subsets, IsaacGym | success rate: 0.54 -> 0.68 | Improved |
| replace 16-layer MLP diffusion backbone with 4-layer DiT (ablation) | RoboGhost motion generator · HumanML tracking and HumanML3D | success rate / FID / time: Succ 0.97 -> 0.96; FID 11.706 -> 11.697; time 5.84 -> 14.28 | No clear effect |
| add causal adaptive sampling to teacher training (ablation) | RoboGhost · Kungfu, IsaacGym | success rate: 0.62 -> 0.72 | Improved |
| add causal adaptive sampling to teacher training (ablation) | RoboGhost · HumanML, IsaacGym | success rate: 0.92 -> 0.97 | Improved |
| vary number of MoE experts in teacher (ablation) | RoboGhost teacher · tracking, IsaacGym/MuJoCo | success rate | No clear effect |
| use RoboGhost latent-driven policy instead of MLP baseline | RoboGhost · Kungfu, MuJoCo | success rate: 0.51 -> 0.57 | Improved |
Closest papers by meaning
- Humanoid Locomotion as Next Token Prediction (NeurIPS 2024 · similarity 0.89)
- RoboMirror: Understand Before You Imitate for Video to Humanoid Locomotion (ECCV 2026 · similarity 0.89)
- End-to-End Language-Action Model for Humanoid Whole Body Control (CVPR 2026 · similarity 0.89)
- Humanoid Horizon: Extending Task Horizon in Whole-Body Loco-Manipulation via Parallel Training, Dynamic Starting, and Reward Gating (NeurIPS 2026 · similarity 0.88)
- SUGAR: A Scalable Human-Video-Driven Generalizable Humanoid Loco-Manipulation Learning Framework (NeurIPS 2026 · similarity 0.86)
- ElegantVLA: Learning When to Think for Efficient Vision-Language-Action Models (NeurIPS 2026 · similarity 0.86)
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.
