E-Ro Nguyen* Yichi Zhang*, Kanchana Ranasinghe, Xiang Li, Michael S Ryoo

Stony Brook University

*Equal contribution

DAWN a unified diffusion-based framework for language-conditioned robotic manipulation that bridges high-level motion intent and low-level robot action via structured pixel motion representation.

• [Feb 2026] 🎉 Our paper has been accepted to CVPR 2026!
• [Sep 2025] 📄 Initial arXiv release.

🚀 Code & pretrained checkpoints will be released soon!

If you find our work useful, please consider citing:

@article{nguyen2025dawn,
  title   = {Pixel Motion Diffusion is What We Need for Robot Control},
  author  = {Nguyen, E-Ro and Zhang, Yichi and Ranasinghe, Kanchana and Li, Xiang and Ryoo, Michael S},
  journal = {arXiv preprint arXiv:2509.22652},
  year    = {2025}
}