Research map / Reinforcement learning / Visual and representation RL
HWC-Loco: A Hierarchical Whole-Body Control Approach to Robust Humanoid Locomotion
Sixu Lin, Guanren Qiao, Yunxin Tai, Ang Li, Kui Jia, Guiliang Liu
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 |
|---|---|---|---|
| add safety recovery policy and high-level planner to goal-tracking policy (ablation) | HWC-Loco · Stairs high-speed | success rate (%): 72.60 -> 84.34 | Improved |
| add safety recovery with high-level policy (ablation) | HWC-Loco · Stairs low-speed | success rate (%): 96.60 -> 99.98 | Improved |
| add safety recovery with high-level policy (ablation) | HWC-Loco · Stairs high-speed | goal tracking: 1.11 -> 1.07 | Hurt |
| use higher alpha (more safety sensitivity) instead of HWC-Loco-l (ablation) | HWC-Loco · Stairs high-speed | success rate (%): 78.92 -> 84.34 | Improved |
| add safety recovery with high-level policy (ablation) | HWC-Loco · Constant force/torque disturbance | success rate (%): 61.20 -> 75.95 | Improved |
| add safety recovery with high-level policy (ablation) | HWC-Loco · High impulse on CoM | success rate (%): 58.13 -> 81.27 | Improved |
| add human imitation objective to DreamWaQ-style policy (HWC-Loco vs DreamWaQ) | HWC-Loco · Stairs high-speed | success rate (%): 60.58 -> 84.34 | Improved |
| increase observation history length from 1 to 10 (ablation) | HWC-Loco · Success rate (history ablation) | success rate (%): 95.13 -> 96.73 | Improved |
| increase observation history length from 10 to 40 (ablation) | HWC-Loco · History ablation | goal-tracking: 1.10 -> 1.02 | Hurt |
| increase alpha to 200 in high-level policy (ablation) | High-level policy · Training | behavior | Hurt |
Closest papers by meaning
- Hierarchical World Models as Visual Whole-Body Humanoid Controllers (ICLR 2025 · similarity 0.89)
- HumanoidArena: Benchmarking Egocentric Hierarchical Whole-body Learning (NeurIPS 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)
- Adversarial Locomotion and Motion Imitation for Humanoid Policy Learning (NeurIPS 2025 · similarity 0.88)
- H-GAP: Humanoid Control with a Generalist Planner (ICLR 2024 · similarity 0.87)
- SUGAR: A Scalable Human-Video-Driven Generalizable Humanoid Loco-Manipulation Learning Framework (NeurIPS 2026 · similarity 0.87)
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.
