Research map / Reinforcement learning / Policy gradient and actor-critic
Mean Flow Policy with Instantaneous Velocity Constraint for One-step Action Generation
Guojian Zhan, Letian Tao, Pengcheng Wang, Yixiao Wang, Yuxin Chen, Yiheng Li, Hongyang Li, Masayoshi Tomizuka, Shengbo Eben Li
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 instantaneous velocity constraint (lambda 0 -> 1.0) (ablation) | MVP · Cube-triple-task4 | success rate: 0.30 -> 0.52 | Improved |
| add instantaneous velocity constraint (lambda 0 -> 1.0) (ablation) | MVP · Cube-triple-task3 | success rate: 0.65 -> 0.71 | Improved |
| replace multi-step flow with mean-velocity one-step policy (ablation) | QC-style chunked best-of-N policy · Cube-triple-task4 | success rate: 0.01 -> 0.52 | Improved |
| replace standard one-step flow with mean velocity policy (ablation) | FQL-Onestep · Cube-triple-task3 | success rate: 0.00 -> 0.71 | Improved |
| use one-step mean velocity generation instead of 10-step flow | BFN/QC · Inference time | ms: 117.3 -> 10.93 ms | Improved |
| use mean velocity policy instead of FQL distillation | FQL · Online training speed | iter/s: 108.5 -> 153.6 iter/s | Improved |
| use mean velocity policy vs QC on easier tasks | QC · Robomimic-can | success rate: 0.94 -> 0.92 | Hurt |
| use MVP vs QC on hardest task | QC · Cube-triple-task4 | success rate: 0.46 -> 0.52 | Improved |
Closest papers by meaning
- Aligning Flow Map Policies with Optimal $Q$-Guidance (NeurIPS 2026 · similarity 0.90)
- Generative Control as Optimization: Time Unconditional Flow Matching for Adaptive and Robust Robotic Control (NeurIPS 2026 · similarity 0.89)
- Flow-Based Policy for Online Reinforcement Learning (NeurIPS 2025 · similarity 0.89)
- FlowPolicy: Enabling Fast and Robust 3D Flow-Based Policy via Consistency Flow Matching for Robot Manipulation (AAAI 2025 · similarity 0.88)
- ReinFlow: Fine-tuning Flow Matching Policy with Online Reinforcement Learning (NeurIPS 2025 · similarity 0.88)
- FreqPolicy: Efficient Flow-based Visuomotor Policy via Frequency Consistency (NeurIPS 2025 · similarity 0.87)
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.
