{"ID":2895005,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2507.10543","arxiv_id":"2507.10543","title":"MP1: MeanFlow Tames Policy Learning in 1-step for Robotic Manipulation","abstract":"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.","short_abstract":"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 a...","url_abs":"https://arxiv.org/abs/2507.10543","url_pdf":"https://arxiv.org/pdf/2507.10543v5","authors":"[\"Juyi Sheng\",\"Ziyi Wang\",\"Peiming Li\",\"Mengyuan Liu\"]","published":"2025-07-14T17:59:08Z","proceeding":"cs.RO","tasks":"[\"cs.RO\"]","methods":"[\"Diffusion Model\"]","has_code":false,"code_links":[{"ID":612148,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2895005,"paper_url":"https://arxiv.org/abs/2507.10543","paper_title":"MP1: MeanFlow Tames Policy Learning in 1-step for Robotic Manipulation","repo_url":"https://github.com/LogSSim/MP1","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}