Research map / Reinforcement learning / Visual and representation RL
Deep Imitation Learning for Bimanual Robotic Manipulation
Fan Xie, Alexander Chowdhury, M. Clara De Paolis Kaluza, Linfeng Zhao, Lawson Wong, Rose Yu
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 |
|---|---|---|---|
| Use hierarchical primitive planning with modular per-primitive dynamics, graph features, and residual links | HDR-IL · Table lifting | success rate: 13% -> 100% | Improved |
| Use hierarchical primitive planning with modular per-primitive dynamics, graph features, and residual links | HDR-IL · Peg-in-hole | lift success rate: 1% -> 29% | Improved |
| Add graph features and a residual link together (ablation) | single-model GRU-GRU · Table lifting | success rate: 13% -> 72% | Improved |
| Add a residual link alone (ablation) | single-model GRU-GRU · Table lifting | success rate: 13% -> 13% | No clear effect |
| Add graph features alone (ablation) | single-model GRU-GRU · Table lifting | success rate: 13% -> 17% | Improved |
| Use modular per-primitive models without graph features or residual links (ablation) | GRU-GRU · Table lifting | success rate: 13% -> 14% | No clear effect |
| Add modular per-primitive models and residual links, without graph features (ablation) | GRU-GRU · Table lifting | success rate: 13% -> 92% | Improved |
| Add modular per-primitive models and graph features, without residual links (ablation) | GRU-GRU · Table lifting | success rate: 13% -> 13% | No clear effect |
| Add modular per-primitive models to ResInt (ablation) | ResInt · Table lifting | success rate: 72% -> 100% | Improved |
| Replace absolute coordinates with relational coordinates (ablation) | residual-link model · Table lifting | success rate: 72% -> 16% | Hurt |
| Add graph features and a residual link (ablation) | GRU-GRU · Peg-in-hole | lift success rate: 1% -> 15% | Improved |
| Use modular per-primitive dynamics in addition to graph features and residual links (ablation) | ResInt · Peg-in-hole | lift success rate: 15% -> 29% | Improved |
Closest papers by meaning
- RDT-1B: a Diffusion Foundation Model for Bimanual Manipulation (ICLR 2025 · similarity 0.91)
- Rethinking Bimanual Robotic Manipulation: Learning with Decoupled Interaction Framework (ICCV 2025 · similarity 0.90)
- Diffusion-Based Imaginative Coordination for Bimanual Manipulation (ICCV 2025 · similarity 0.89)
- AnyBimanual: Transferring Unimanual Policy for General Bimanual Manipulation (ICCV 2025 · similarity 0.89)
- Spatial-Temporal Graph Diffusion Policy with Kinematic Modeling for Bimanual Robotic Manipulation (CVPR 2025 · similarity 0.88)
- ManipTrans: Efficient Dexterous Bimanual Manipulation Transfer via Residual Learning (CVPR 2025 · 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.
