# MP1: MeanFlow Tames Policy Learning in 1-step for Robotic Manipulation

arXiv: https://arxiv.org/abs/2507.10543 · HTML: https://mlsanity.com/paper/2507.10543/

## Key facts

- Cited by 35 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 14 Jul 2025.

## Built on by (most cited first, 30 of 35)

- [Much Ado About Noising: Dispelling the Myths of Generative Robotic Control](https://mlsanity.com/paper/2512.01809/) (2025)
- [VITA: Vision-to-Action Flow Matching Policy](https://mlsanity.com/paper/2507.13231/) (2025)
- [One-Step Flow Policy Mirror Descent](https://mlsanity.com/paper/2507.23675/) (2025)
- [FASTER: Rethinking Real-Time Flow VLAs](https://mlsanity.com/paper/2603.19199/) (2026)
- [One-Step Generative Policies with Q-Learning: A Reformulation of MeanFlow](https://mlsanity.com/paper/2511.13035/) (2025)
- [MeanFuser: Fast One-Step Multi-Modal Trajectory Generation and Adaptive Reconstruction via MeanFlow for End-to-End Autonomous Driving](https://mlsanity.com/paper/2602.20060/) (2026)
- [DM1: MeanFlow with Dispersive Regularization for 1-Step Robotic Manipulation](https://mlsanity.com/paper/2510.07865/) (2025)
- [STEP: Warm-Started Visuomotor Policies with Spatiotemporal Consistency Prediction](https://mlsanity.com/paper/2602.08245/) (2026)
- [Mean-Flow based One-Step Vision-Language-Action](https://mlsanity.com/paper/2603.01469/) (2026)
- [One-Step Flow Policy: Self-Distillation for Fast Visuomotor Policies](https://mlsanity.com/paper/2603.12480/) (2026)
- [Drift-Based Policy Optimization: Native One-Step Policy Learning for Online Robot Control](https://mlsanity.com/paper/2604.03540/) (2026)
- [EfficientFlow: Efficient Equivariant Flow Policy Learning for Embodied AI](https://mlsanity.com/paper/2512.02020/) (2025)
- [OMP: One-step Meanflow Policy with Directional Alignment](https://mlsanity.com/paper/2512.19347/) (2025)
- [HeRO: Hierarchical 3D Semantic Representation for Pose-aware Object Manipulation](https://mlsanity.com/paper/2602.18817/) (2026)
- [Hyper-DP3: Frequency-Aware Right-Sizing of 3D Diffusion Policies for Visuomotor Control](https://mlsanity.com/paper/2605.01581/) (2026)
- [Drifting Field Policy: A One-Step Generative Policy via Wasserstein Gradient Flow](https://mlsanity.com/paper/2605.07727/) (2026)
- [FA-RDP: A Frequency-Adaptive Reactive Diffusion Policy for Contact-Rich Manipulation](https://mlsanity.com/paper/2607.28596/) (2026)
- [Latent Policy Steering through One-Step Flow Policies](https://mlsanity.com/paper/2603.05296/) (2026)
- [Ada3Drift: Adaptive Training-Time Drifting for One-Step 3D Visuomotor Robotic Manipulation](https://mlsanity.com/paper/2603.11984/) (2026)
- [Drift Flow Matching](https://mlsanity.com/paper/2605.17244/) (2026)
- [Where Should Action Generation Begin? A Learnable Source Prior for Generative Robot Policies](https://mlsanity.com/paper/2606.17408/) (2026)
- [DriftingVLA: Native One-Step Vision-Language-Action Generation via Per-Dimension Temporal Drifting](https://mlsanity.com/paper/2608.29749/) (2026)
- [Decoupling Vision, Language, and Action for Efficient Multi-Task Robot Policies](https://mlsanity.com/paper/2609.18374/) (2026)
- [DEFLECT: Temporal Counterfactual Preference Learning for Delay-Robust Asynchronous VLAs](https://mlsanity.com/paper/2605.19294/) (2026)
- [Implicit Drifting Policy: One-Step Action Generation via Conditional Expert Geometry](https://mlsanity.com/paper/2606.01098/) (2026)
- [High-Fidelity One-Step Generative Visuomotor Policy via Recursive Correction, Frequency Consistency, and Contrastive Flow Matching](https://mlsanity.com/paper/2607.03865/) (2026)
- [SUN: Agentic Robot Policy Learning with Persistent Task Programs](https://mlsanity.com/paper/2608.31167/) (2026)
- [KAN We Flow? Advancing Robotic Manipulation with 3D Flow Matching via KAN & RWKV](https://mlsanity.com/paper/2602.01115/) (2026)
- [HybridFlow: A 2-NFE Generative Policy for Real-Time Robotic Manipulation](https://mlsanity.com/paper/2602.13718/) (2026)
- [Discrete MeanFlow: One-Step Generation via Conditional Transition Kernels](https://mlsanity.com/paper/2605.12805/) (2026)

## Abstract (arXiv)

In robot manipulation, robot learning has become a prevailing approach. However, generative models within this field face a fundamental trade-off between the slow, iterative sampling of diffusion models and the architectural constraints of faster Flow-based methods, which often rely on explicit consistency losses. To address these limitations, we introduce MP1, which pairs 3D point-cloud inputs with the MeanFlow paradigm to generate action trajectories in one network function evaluation (1-NFE). By directly learning the interval-averaged velocity via the "MeanFlow Identity", our policy avoids any additional consistency constraints. This formulation eliminates numerical ODE-solver errors during inference, yielding more precise trajectories. MP1 further incorporates CFG for improved trajectory controllability while retaining 1-NFE inference without reintroducing structural constraints. Because subtle scene-context variations are critical for robot learning, especially in few-shot learning, we introduce a lightweight Dispersive Loss that repels state embeddings during training, boosting generalization without slowing inference. We validate our method on the Adroit and Meta-World benchmarks, as well as in real-world scenarios. Experimental results show MP1 achieves superior average task success rates, outperforming DP3 by 10.2% and FlowPolicy by 7.3%. Its average inference time is only 6.8 ms-19x faster than DP3 and nearly 2x faster than FlowPolicy. Our project page is available at https://mp1-2254.github.io/, and the code can be accessed at https://github.com/LogSSim/MP1.