Research map / Video / Action recognition
Diffusion-Based Imaginative Coordination for Bimanual Manipulation
Huilin Xu, Jian Ding, Jiakun Xu, Ruixiang Wang, Jun Chen, Jinjie Mai, Yanwei Fu, Bernard Ghanem, Feng Xu, Mohamed Elhoseiny
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 video prediction as auxiliary task (ablation) | Ours · ALOHA (sim) | avg success rate (%): 64.6 -> 71.9 | Improved |
| predict latents instead of pixels (next frame) (ablation) | Ours · ALOHA (sim) | avg success rate (%): 68.7 -> 71.3 | Improved |
| predict multiple frames instead of next frame (latents) (ablation) | Ours · ALOHA (sim) | avg success rate (%): 71.3 -> 71.9 | Improved |
| use unidirectional instead of full attention (ablation) | Ours · ALOHA (sim) | avg success rate (%): 70.8 -> 71.9 | Improved |
| use unidirectional instead of decoupled attention (ablation) | Ours · ALOHA (sim) | avg success rate (%): 70.1 -> 71.9 | Improved |
| unidirectional attention lets video prediction be skipped at inference (ablation) | Ours · ALOHA (sim) | inference time (s): 0.06 -> 0.03 | Improved |
| set video prediction weight to 0.2 rather than 0.05 (ablation) | Ours · ALOHA (sim) | avg success rate (%): 71.1 -> 71.9 | Improved |
| set video prediction weight to 0.2 rather than 0.5 (ablation) | Ours · ALOHA (sim) | avg success rate (%): 70.6 -> 71.9 | Improved |
| predict 40 future frames rather than 20 (ablation) | Ours · ALOHA (sim) | avg success rate (%): 71.1 -> 71.9 | Improved |
| predict 40 future frames rather than 100 (ablation) | Ours · ALOHA (sim) | avg success rate (%): 70.6 -> 71.9 | Improved |
| use discrete instead of continuous video tokens (ablation) | Ours · ALOHA (sim) (video token type) | avg success rate (%): 70.7 -> 73.7 | Improved |
| add video prediction on Seq-coordinated tasks (ablation) | Ours · RoboTwin Seq-Coordinated tasks | success rate (%): +6.2 | Improved |
Closest papers by meaning
- RDT-1B: a Diffusion Foundation Model for Bimanual Manipulation (ICLR 2025 · similarity 0.91)
- Robots Pre-train Robots: Manipulation-Centric Robotic Representation from Large-Scale Robot Datasets (ICLR 2025 · similarity 0.89)
- VidMan: Exploiting Implicit Dynamics from Video Diffusion Model for Effective Robot Manipulation (NeurIPS 2024 · similarity 0.89)
- Video Prediction Policy: A Generalist Robot Policy with Predictive Visual Representations (ICML 2025 · similarity 0.88)
- Spatial-Temporal Graph Diffusion Policy with Kinematic Modeling for Bimanual Robotic Manipulation (CVPR 2025 · similarity 0.88)
- VideoVLA: Video Generators Can Be Generalizable Robot Manipulators (NeurIPS 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.
