Research map / Reinforcement learning / Imitation learning
Hierarchical Value-Decomposed Offline Reinforcement Learning for Whole-Body Control
Zhilong Zhang, Yunpeng Mei, Xinghao Du, Hongjie Cao, Haonan Wang, Pengyuan Min, Chenyu Wang, Pengfei Chen, Chenbo Xin, Yijie Wang, Wenyu Luo, Yihao Sun and 4 more
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 hierarchical Q-value decomposition (vs shared Q-value) (ablation) | HVD with DP · WB-50, average over 5 tasks | success rate: -0.03 | Improved |
| add hierarchical Q-value decomposition (vs shared Q-value) (ablation) | HVD with WB-VIMA · WB-50, average over 5 tasks | success rate: -0.07 | Improved |
| add hierarchical Q-value decomposition (vs shared Q-value) (ablation) | HVD with π0 · WB-50, average over 5 tasks | success rate: -0.05 | Improved |
| add hierarchical Q-value decomposition (vs shared Q-value) (ablation) | HVD with π0 · WB-50 Pen Insert | success rate: +0.04 | Hurt |
| add hierarchical Q-value decomposition (vs shared Q-value) (ablation) | HVD with π0 · WB-50 Wipe Board | success rate: -0.14 | Improved |
| add hierarchical Q-value decomposition (vs shared Q-value) (ablation) | HVD with WB-VIMA · WB-50 Wipe Board | success rate: -0.12 | Improved |
| train HVD on mixed-quality data instead of IL on expert-only data | π0 · WB-50, average over 5 tasks | success rate: 0.44 -> 0.58 | Improved |
| train HVD on mixed-quality data instead of IL on expert-only data | WB-VIMA · WB-50, average over 5 tasks | success rate: 0.30 -> 0.46 | Improved |
| train HVD on mixed-quality data instead of IL on expert-only data | Diffusion Policy · WB-50, average over 5 tasks | success rate: 0.26 -> 0.32 | Improved |
| train HVD on mixed-quality data instead of IL on expert-only data | WB-VIMA · WB-50 Basket Carry | success rate: 0.10 -> 0.10 | No clear effect |
| train HVD on mixed-quality data instead of IL on expert-only data | Diffusion Policy · WB-50 Wipe Board | success rate: 0.00 -> 0.00 | No clear effect |
| train one multi-task policy instead of per-task specialists | π0 IL · WB-50, average over 5 tasks | success rate: 0.44 -> 0.36 | Hurt |
Closest papers by meaning
- Hierarchical Equivariant Policy via Frame Transfer (ICML 2025 · similarity 0.88)
- Hierarchical World Models as Visual Whole-Body Humanoid Controllers (ICLR 2025 · similarity 0.88)
- GO-DICE: Goal-Conditioned Option-Aware Offline Imitation Learning via Stationary Distribution Correction Estimation (AAAI 2024 · similarity 0.88)
- Adversarial Locomotion and Motion Imitation for Humanoid Policy Learning (NeurIPS 2025 · similarity 0.87)
- HumanoidArena: Benchmarking Egocentric Hierarchical Whole-body Learning (NeurIPS 2026 · similarity 0.87)
- Offline Actor-Critic Reinforcement Learning Scales to Large Models (ICML 2024 · 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.
