Research map / Vision-language and multimodal / Robot action and navigation
Humanoid Whole-Body Manipulation via Active Spatial Brain and Generalizable Action Cerebellum
Zhizhao Liang, Yi-Lin Wei, Xuhang Chen, Mu Lin, Yi-Xiang He, Zhexi Luo, Jun-Hui Liu, Kun-Yu Lin, WEISHI ZHENG
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 Active Perception (ablation) | Ours (no AP, no EES) · Different Heights and Positions | success rate (%): 18.8 -> 55.6 | Improved |
| add End-Effector Reachable Space Solver (ablation) | Ours (no AP, no EES) · Different Heights and Positions | success rate (%): 18.8 -> 24.9 | Improved |
| add both AP and EES (ablation) | Ours · Different Heights and Positions | success rate (%): 18.8 -> 67.5 | Improved |
| add Active Perception (ablation) | Ours (no AP, no EES) · Obstacle Avoidance | success rate (%): 0 -> 16.7 | Improved |
| add End-Effector Reachable Space Solver (ablation) | Ours (no AP, no EES) · Obstacle Avoidance | success rate (%): 0 -> 33.4 | Improved |
| add both AP and EES (ablation) | Ours · Obstacle Avoidance | success rate (%): 0 -> 70.0 | Improved |
| use multi-agent VLM framework instead of data-driven Psi0 on hard tasks | Ours · Obstacle Avoidance (hard) | success rate (%): 0 -> 60.0 | Improved |
| use data-driven TrajBooster on hard instead of easy setting | TrajBooster · Occlusion Handling | success rate (%): 40.0 -> 20.0 | Hurt |
| use unseen tools object category for our framework | Ours · Generalization across objects | success rate (%): 77.8 -> 52.0 | Hurt |
| use unseen items category for our framework | Ours · Generalization across objects | success rate (%): 60.0 -> 69.6 | Improved |
| use our framework instead of TrajBooster on easy Different Heights | Ours · Different Heights (easy) | success rate (%): 70.0 -> 85.0 | Improved |
| use gemini-3.1-pro instead of gpt5.1 for waypoint generation | VLM benchmark · Waypoints generation | Appropriate (%): 30.35 -> 64.91 | Improved |
Closest papers by meaning
- Humanoid Horizon: Extending Task Horizon in Whole-Body Loco-Manipulation via Parallel Training, Dynamic Starting, and Reward Gating (NeurIPS 2026 · similarity 0.89)
- RoboBrain: A Unified Brain Model for Robotic Manipulation from Abstract to Concrete (CVPR 2025 · similarity 0.86)
- Hierarchical World Models as Visual Whole-Body Humanoid Controllers (ICLR 2025 · similarity 0.86)
- SUGAR: A Scalable Human-Video-Driven Generalizable Humanoid Loco-Manipulation Learning Framework (NeurIPS 2026 · similarity 0.86)
- SPIN: Simultaneous Perception Interaction and Navigation (CVPR 2024 · similarity 0.86)
- R2C: Mapping Room to Chessboard to Unlock LLM As Low-Level Action Planner (CVPR 2025 · 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.
