# OGPO: One-Step Generative Policy Optimization for Real-Time Robot Control

arXiv: https://arxiv.org/abs/2601.20701 · HTML: https://mlsanity.com/paper/2601.20701/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 28 Jan 2026.

## Built on by (most cited first, 7 of 7)

- [Drift-Based Policy Optimization: Native One-Step Policy Learning for Online Robot Control](https://mlsanity.com/paper/2604.03540/) (2026)
- [Advances in GRPO for Generation Models: A Survey](https://mlsanity.com/paper/2603.06623/) (2026)
- [Ada3Drift: Adaptive Training-Time Drifting for One-Step 3D Visuomotor Robotic Manipulation](https://mlsanity.com/paper/2603.11984/) (2026)
- [CoFlow: Coordinated Few-Step Flow for Offline Multi-Agent Decision Making](https://mlsanity.com/paper/2605.01457/) (2026)
- [HybridFlow: A 2-NFE Generative Policy for Real-Time Robotic Manipulation](https://mlsanity.com/paper/2602.13718/) (2026)
- [NavCMPO: Critic-Guided MeanFlow Policy Optimization for Adaptive Navigation](https://mlsanity.com/paper/2607.14643/) (2026)
- [MA-FPPO: Multi-Agent Flow-Pretrained Policy Optimization](https://mlsanity.com/paper/2609.32594/) (2026)

## Abstract (arXiv)

Real-time robot control demands fast action generation. Diffusion and flow matching policies for robot control require multi-step sampling, limiting their deployment in real-time scenarios. Natively reducing the sampling steps to one sacrifices representation quality and task performance, creating a trilemma among speed, fidelity, and performance. We present One-Step Generative Policy Optimization (OGPO), a systematic framework to resolve this trilemma. OGPO first pairs a lightweight architecture with the interval velocity principle for distillation-free one-step inference, while representation spreading prevents representation quality degradation. It then performs on-policy reinforcement learning (RL) fine-tuning on this fast, stable policy to break the imitation learning ceiling. Experiments on RoboMimic and OpenAI Gym benchmarks show that OGPO matches or exceeds multi-step baselines while achieving a 5-20 times inference speedup and over 120Hz control frequency. Physical deployment on a Franka-Emika-Panda robot validates real-world applicability. Project page: https://ogpo-project.github.io/