Research map / Vision-language and multimodal / Robot action and navigation
H-RDT: Human Manipulation Enhanced Bimanual Robotic Manipulation
Hongzhe Bi, Lingxuan Wu, Tianwei Lin, Hengkai Tan, Zhizhong Su, Hang Su, Jun Zhu
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 human egocentric pre-training (EgoDex) (ablation) | H-RDT · RoboTwin 2.0 multi-task (Hard mode) | average success rate (%): 67.2% -> 87.2% | Improved |
| add human pre-training (ablation) | H-RDT · RoboTwin 2.0 single-task (13 tasks) | average success rate (%): +8.4% in both Easy and Hard modes | Improved |
| add human pre-training (ablation) | H-RDT · Real-world few-shot ARX5 (113 tasks) | average success rate (%): 17.6% -> 41.6% | Improved |
| add human pre-training (ablation) | H-RDT · Real-world dual UR5 + UMI | average success rate (%): 16.0% -> 58.0% | Improved |
| add human pre-training (ablation) | H-RDT · Real-world towel folding | complete success rate: 0% -> 52% | Improved |
| add human pre-training (ablation) | H-RDT · Real-world cup to coaster | complete success rate: 20% -> 64% | Improved |
| use H-RDT instead of pi0 | few-shot ARX5 policy · Real-world few-shot ARX5 | average success rate (%): 31.2% -> 41.6% | Improved |
| use H-RDT instead of pi0 on RoboTwin multi-task | bimanual policy · RoboTwin 2.0 multi-task (Hard mode) | average success rate (%): 48.4% -> 87.2% | Improved |
| human pre-training benefit on multi-task versus single-task (ablation) | H-RDT · RoboTwin 2.0 | improvement over scratch: +20.0% multi-task vs +8.4% single-task | Improved |
Closest papers by meaning
- RDT-1B: a Diffusion Foundation Model for Bimanual Manipulation (ICLR 2025 · similarity 0.94)
- Robots Pre-train Robots: Manipulation-Centric Robotic Representation from Large-Scale Robot Datasets (ICLR 2025 · similarity 0.90)
- ManipTrans: Efficient Dexterous Bimanual Manipulation Transfer via Residual Learning (CVPR 2025 · similarity 0.89)
- Mitigating the Human-Robot Domain Discrepancy in Visual Pre-training for Robotic Manipulation (CVPR 2025 · similarity 0.89)
- VidBot: Learning Generalizable 3D Actions from In-the-Wild 2D Human Videos for Zero-Shot Robotic Manipulation (CVPR 2025 · similarity 0.88)
- HandEdit: A Unified Benchmark for Egocentric Human-to-Robot Dexterous Hand Image Editing (NeurIPS 2026 · similarity 0.88)
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.
