OMP: One-step Meanflow Policy with Directional Alignment

Han Fang1, Yize Huang2, Yuheng Zhao1, Paul Weng3, Xiao Li2, Yutong Ban1
1 Global College, Shanghai Jiao Tong University
2 School of Mechanical Engineering, Shanghai Jiao Tong University
3 Duke Kunshan University
OMP overview figure

OMP introduces a directional alignment mechanism and a memory-efficient Differential Derivation Equation (DDE) to achieve accurate, single-step policy generation for robot manipulation.

Abstract

Robot manipulation has increasingly adopted data-driven generative policy frameworks, yet the field faces a persistent trade-off: diffusion models suffer from high inference latency, while flow-based methods often require complex architectural constraints. Although in image generation domain, the MeanFlow paradigm offers a path to single-step inference, its direct application to robotics is impeded by critical theoretical pathologies, specifically spectral bias and gradient starvation in low-velocity regimes. To overcome these limitations, we propose the One-step MeanFlow Policy (OMP), a novel framework designed for high-fidelity, real-time manipulation. We introduce a lightweight directional alignment mechanism to explicitly synchronize predicted velocities with true mean velocities. Furthermore, we implement a Differential Derivation Equation (DDE) to approximate the Jacobian-Vector Product (JVP) operator, which decouples forward and backward passes to significantly reduce memory complexity. Extensive experiments on the Adroit and Meta-World benchmarks demonstrate that OMP outperforms state-of-the-art methods in success rate and trajectory accuracy, particularly in high-precision tasks, while retaining the efficiency of single-step generation.

Experimental Results

OMP benchmark results

Quantitative and qualitative comparisons on Adroit, Meta-World, and real-world tasks show strong performance, robust training behavior, and real-time inference capability.

Real-World Robot Demonstration

Real-robot demonstrations on representative manipulation tasks, including bottle placement, table cleaning, and slip-ring insertion.

BibTeX

@article{fang2026omp,
  title={OMP: One-step Meanflow Policy with Directional Alignment},
  author={Han Fang and Yize Huang and Yuheng Zhao and Paul Weng and Xiao Li and Yutong Ban},
  journal={arXiv preprint arXiv:2512.19347},
  year={2026},
  url={https://arxiv.org/abs/2512.19347}
}