@inproceedings {pub6174,
	title = {Fast Flow-based Visuomotor Policies via Conditional Optimal Transport Couplings},
	author = {Andreas Sochopoulos AND Nikolaos Tsagkas AND Joao Moura AND Nikolay Malkin AND Michael Gienger AND Sethu Vijayakumar},
	year = {2025},
	month = {September},
	abstract = {Diffusion and flow matching policies have recently shown remarkable performance in robotic applications by accurately capturing multimodal robot trajectory distributions. However, their computationally expensive inference, due to numerical integration of an ODE or SDE, limits their application as real-time controllers for robots. We introduce a methodology that utilizes conditional Optimal Transport couplings between noise and samples, in order to force straight solutions on the Probability Flow ODE in robot action generation tasks. We demonstrate that naively coupling noise and samples fails in conditional tasks and introduce the condition variables in the coupling process to enhance few-step performance. The proposed few-step policy achieves $4\%$ higher success rate with a 10x speed-up compared to Diffusion Policy, in a diverse set of simulation tasks. Our method also maintains the same training complexity as Diffusion Policy and vanilla Flow Matching, in contrast to distillation methods.},
	publisher = {Proceedings of Machine Learning Research (PMLR)},
	url = {https://proceedings.mlr.press/v305/sochopoulos25a.html},
	booktitle = {Conference on Robot Learning}
}