Research map / Image generation / Fast diffusion sampling
OMP: One-step Meanflow Policy with Directional Alignment
Han Fang, Yize Huang, Yuheng Zhao, Paul Weng, Xiao Li, Yutong Ban
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add directional alignment loss (ablation) | OMP-JVP · Adroit + Meta-World average | success rate: 78.9 -> 82.3 | Improved |
| add directional alignment loss (ablation) | OMP-DDE · Adroit + Meta-World average | success rate: 77.2 -> 80.8 | Improved |
| add dispersive loss (ablation) | OMP-JVP · Adroit + Meta-World average | success rate: 81.2 -> 82.3 | Improved |
| add dispersive loss (ablation) | OMP-DDE · Adroit + Meta-World average | success rate: 80.1 -> 80.8 | Improved |
| add both dispersive and directional alignment losses (ablation) | OMP-JVP · Adroit Pen | success rate: 48 -> 60 | Improved |
| add both dispersive and directional alignment losses (ablation) | OMP-DDE · Adroit Pen | success rate: 55 -> 64 | Improved |
| replace exact JVP with finite-difference DDE | OMP · Adroit + Meta-World average | success rate: 82.3 -> 80.8 | Hurt |
| replace exact JVP with finite-difference DDE | OMP · Meta-World Very Hard | success rate: 77.8 -> 70.6 | Hurt |
| replace exact JVP with finite-difference DDE | OMP · Place Bottle (Horizon=16) GPU memory | GPU memory (GB): 26.71 -> 19.19 | Improved |
| replace exact JVP with finite-difference DDE | OMP · Adroit Hammer (Horizon=4) GPU memory | GPU memory (GB): 6.60 -> 5.35 | Improved |
| replace MP1 with OMP (directional alignment) | OMP-JVP · Meta-World Medium | success rate: 68.0 -> 77.4 | Improved |
| replace MP1 with OMP (directional alignment) | OMP-JVP · Adroit Door | success rate: 69 -> 68 | No clear effect |
Closest papers by meaning
- FlowPolicy: Enabling Fast and Robust 3D Flow-Based Policy via Consistency Flow Matching for Robot Manipulation (AAAI 2025 · similarity 0.89)
- FreqPolicy: Efficient Flow-based Visuomotor Policy via Frequency Consistency (NeurIPS 2025 · similarity 0.88)
- Generative Control as Optimization: Time Unconditional Flow Matching for Adaptive and Robust Robotic Control (NeurIPS 2026 · similarity 0.87)
- Implicit Drifting Policy: One-Step Action Generation via Conditional Expert Geometry (NeurIPS 2026 · similarity 0.86)
- Mean Flows for One-step Generative Modeling (NeurIPS 2025 · similarity 0.86)
- FlowRAM: Grounding Flow Matching Policy with Region-Aware Mamba Framework for Robotic Manipulation (CVPR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
