One-Step Diffusion Policy: Fast Visuomotor Policies via Diffusion Distillation

Zhendong Wang, Max Li, Ajay Mandlekar, Zhenjia Xu, Jiaojiao Fan, Yashraj Narang, Linxi Fan, Yuke Zhu, Yogesh Balaji, Mingyuan Zhou, Ming-Yu Liu, Yu Zeng
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:63399-63416, 2025.

Abstract

Diffusion models, praised for their success in generative tasks, are increasingly being applied to robotics, demonstrating exceptional performance in behavior cloning. However, their slow generation process stemming from iterative denoising steps poses a challenge for real-time applications in resource-constrained robotics setups and dynamically changing environments. In this paper, we introduce the One-Step Diffusion Policy (OneDP), a novel approach that distills knowledge from pre-trained diffusion policies into a single-step action generator, significantly accelerating response times for robotic control tasks. We ensure the distilled generator closely aligns with the original policy distribution by minimizing the Kullback-Leibler (KL) divergence along the diffusion chain, requiring only $2%$-$10%$ additional pre-training cost for convergence. We evaluated OneDP on 6 challenging simulation tasks as well as 4 self-designed real-world tasks using the Franka robot. The results demonstrate that OneDP not only achieves state-of-the-art success rates but also delivers an order-of-magnitude improvement in inference speed, boosting action prediction frequency from 1.5 Hz to 62 Hz, establishing its potential for dynamic and computationally constrained robotic applications. A video demo is provided at our project page, and the code will be publicly available.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-wang25ba, title = {One-Step Diffusion Policy: Fast Visuomotor Policies via Diffusion Distillation}, author = {Wang, Zhendong and Li, Max and Mandlekar, Ajay and Xu, Zhenjia and Fan, Jiaojiao and Narang, Yashraj and Fan, Linxi and Zhu, Yuke and Balaji, Yogesh and Zhou, Mingyuan and Liu, Ming-Yu and Zeng, Yu}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {63399--63416}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/wang25ba/wang25ba.pdf}, url = {https://proceedings.mlr.press/v267/wang25ba.html}, abstract = {Diffusion models, praised for their success in generative tasks, are increasingly being applied to robotics, demonstrating exceptional performance in behavior cloning. However, their slow generation process stemming from iterative denoising steps poses a challenge for real-time applications in resource-constrained robotics setups and dynamically changing environments. In this paper, we introduce the One-Step Diffusion Policy (OneDP), a novel approach that distills knowledge from pre-trained diffusion policies into a single-step action generator, significantly accelerating response times for robotic control tasks. We ensure the distilled generator closely aligns with the original policy distribution by minimizing the Kullback-Leibler (KL) divergence along the diffusion chain, requiring only $2%$-$10%$ additional pre-training cost for convergence. We evaluated OneDP on 6 challenging simulation tasks as well as 4 self-designed real-world tasks using the Franka robot. The results demonstrate that OneDP not only achieves state-of-the-art success rates but also delivers an order-of-magnitude improvement in inference speed, boosting action prediction frequency from 1.5 Hz to 62 Hz, establishing its potential for dynamic and computationally constrained robotic applications. A video demo is provided at our project page, and the code will be publicly available.} }
Endnote
%0 Conference Paper %T One-Step Diffusion Policy: Fast Visuomotor Policies via Diffusion Distillation %A Zhendong Wang %A Max Li %A Ajay Mandlekar %A Zhenjia Xu %A Jiaojiao Fan %A Yashraj Narang %A Linxi Fan %A Yuke Zhu %A Yogesh Balaji %A Mingyuan Zhou %A Ming-Yu Liu %A Yu Zeng %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-wang25ba %I PMLR %P 63399--63416 %U https://proceedings.mlr.press/v267/wang25ba.html %V 267 %X Diffusion models, praised for their success in generative tasks, are increasingly being applied to robotics, demonstrating exceptional performance in behavior cloning. However, their slow generation process stemming from iterative denoising steps poses a challenge for real-time applications in resource-constrained robotics setups and dynamically changing environments. In this paper, we introduce the One-Step Diffusion Policy (OneDP), a novel approach that distills knowledge from pre-trained diffusion policies into a single-step action generator, significantly accelerating response times for robotic control tasks. We ensure the distilled generator closely aligns with the original policy distribution by minimizing the Kullback-Leibler (KL) divergence along the diffusion chain, requiring only $2%$-$10%$ additional pre-training cost for convergence. We evaluated OneDP on 6 challenging simulation tasks as well as 4 self-designed real-world tasks using the Franka robot. The results demonstrate that OneDP not only achieves state-of-the-art success rates but also delivers an order-of-magnitude improvement in inference speed, boosting action prediction frequency from 1.5 Hz to 62 Hz, establishing its potential for dynamic and computationally constrained robotic applications. A video demo is provided at our project page, and the code will be publicly available.
APA
Wang, Z., Li, M., Mandlekar, A., Xu, Z., Fan, J., Narang, Y., Fan, L., Zhu, Y., Balaji, Y., Zhou, M., Liu, M. & Zeng, Y.. (2025). One-Step Diffusion Policy: Fast Visuomotor Policies via Diffusion Distillation. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:63399-63416 Available from https://proceedings.mlr.press/v267/wang25ba.html.

Related Material