Research map / Reinforcement learning / Visual and representation RL
Towards Human-Level Bimanual Dexterous Manipulation with Reinforcement Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 multi-agent rather than single-agent training for tasks requiring more bimanual cooperation | Bi-DexHands RL policies · More difficult cooperative manipulation tasks | task score | Helped |
| Use PPO instead of HAPPO or MAPPO | Bi-DexHands policies · Most individual tasks | task score | Helped |
| Use SAC rather than on-policy PPO-type training | Bi-DexHands policies · Almost all individual tasks | task score | Hurt |
| Use BCQ or TD3+BC instead of behavior cloning | Offline policies · Hand Over and Door Open Outward | normalized score | Helped |
| Use IQL instead of behavior cloning | Offline policies · Bi-DexHands offline datasets | normalized score | Mixed |
| Train one multi-task policy rather than separate task policies | Multi-task PPO · MT1, MT4, and MT20 | average reward | Hurt |
| Use ProMP few-shot meta-learning rather than random policy | Meta-RL policies · ML1, ML4, and ML20 | average reward | Helped |
| Increase the human-age-associated task difficulty | Bi-DexHands task suite · MT20 tasks ordered by corresponding human age | normalized reward | Hurt |
Closest papers by meaning
- ManipTrans: Efficient Dexterous Bimanual Manipulation Transfer via Residual Learning (CVPR 2025 · similarity 0.91)
- Cross-Embodiment Dexterous Grasping with Reinforcement Learning (ICLR 2025 · similarity 0.88)
- Scaffolding Dexterous Manipulation with Vision-Language Models (NeurIPS 2025 · similarity 0.88)
- HumanoidGen: Data Generation for Bimanual Dexterous Manipulation via LLM Reasoning (NeurIPS 2025 · similarity 0.88)
- DexFlyWheel: A Scalable and Self-improving Data Generation Framework for Dexterous Manipulation (NeurIPS 2025 · similarity 0.87)
- RDT-1B: a Diffusion Foundation Model for Bimanual Manipulation (ICLR 2025 · 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.