# CF-VLA: Efficient Coarse-to-Fine Action Generation for Vision-Language-Action Policies arXiv: https://arxiv.org/abs/2604.24622 · HTML: https://mlsanity.com/paper/2604.24622/ ## Key facts - Cited by 4 papers in the map (data through 14 Oct 2026). - Placed in Efficient VLA inference and adaptation, one of 13 directions in Vision-language-action models. - Builds on 42 of its 55 references that are in the map. - First posted on arXiv 27 Apr 2026. ## Placed in - Vision-language-action models / [Efficient VLA inference and adaptation](https://mlsanity.com/vla/efficient-vla-inference-and-adaptation/) ## Builds on (most cited first, 30 of 42) - [Neural Ordinary Differential Equations](https://mlsanity.com/paper/1806.07366/) (2018) - [RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control](https://mlsanity.com/paper/2307.15818/) (2023) - [OpenVLA: An Open-Source Vision-Language-Action Model](https://mlsanity.com/paper/2406.09246/) (2024) - [Diffusion Policy: Visuomotor Policy Learning via Action Diffusion](https://mlsanity.com/paper/2303.04137/) (2023) - [π0: A Vision-Language-Action Flow Model for General Robot Control](https://mlsanity.com/paper/2410.24164/) (2024) - [DPM-Solver: A Fast ODE Solver for Diffusion Probabilistic Model Sampling in Around 10 Steps](https://mlsanity.com/paper/2206.00927/) (2022) - [π0.5: a Vision-Language-Action Model with Open-World Generalization](https://mlsanity.com/paper/2504.16054/) (2025) - [LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learning](https://mlsanity.com/paper/2306.03310/) (2023) - [CALVIN: A Benchmark for Language-Conditioned Policy Learning for Long-Horizon Robot Manipulation Tasks](https://mlsanity.com/paper/2112.03227/) (2021) - [UniVLA: Learning to Act Anywhere with Task-centric Latent Actions](https://mlsanity.com/paper/2505.06111/) (2025) - [X-VLA: Soft-Prompted Transformer as Scalable Cross-Embodiment Vision-Language-Action Model](https://mlsanity.com/paper/2510.10274/) (2025) - [Cosmos Policy: Fine-Tuning Video Models for Visuomotor Control and Planning](https://mlsanity.com/paper/2601.16163/) (2026) - [DreamVLA: A Vision-Language-Action Model Dreamed with Comprehensive World Knowledge](https://mlsanity.com/paper/2507.04447/) (2025) - [MolmoAct: Action Reasoning Models that can Reason in Space](https://mlsanity.com/paper/2508.07917/) (2025) - [MemoryVLA: Perceptual-Cognitive Memory in Vision-Language-Action Models for Robotic Manipulation](https://mlsanity.com/paper/2508.19236/) (2025) - [What Matters in Language Conditioned Robotic Imitation Learning over Unstructured Data](https://mlsanity.com/paper/2204.06252/) (2022) - [Multimodal Diffusion Transformer: Learning Versatile Behavior from Multimodal Goals](https://mlsanity.com/paper/2407.05996/) (2024) - [DeeR-VLA: Dynamic Inference of Multimodal Large Language Models for Efficient Robot Execution](https://mlsanity.com/paper/2411.02359/) (2024) - [ABot-M0: VLA Foundation Model for Robotic Manipulation with Action Manifold Learning](https://mlsanity.com/paper/2602.11236/) (2026) - [Grounding Language with Visual Affordances over Unstructured Data](https://mlsanity.com/paper/2210.01911/) (2022) - [VLA-Cache: Efficient Vision-Language-Action Manipulation via Adaptive Token Caching](https://mlsanity.com/paper/2502.02175/) (2025) - [Vision-Language-Action Models for Robotics: A Review Towards Real-World Applications](https://mlsanity.com/paper/2510.07077/) (2025) - [Universal Actions for Enhanced Embodied Foundation Models](https://mlsanity.com/paper/2501.10105/) (2025) - [InstructVLA: Vision-Language-Action Instruction Tuning from Understanding to Manipulation](https://mlsanity.com/paper/2507.17520/) (2025) - [Large VLM-based Vision-Language-Action Models for Robotic Manipulation: A Survey](https://mlsanity.com/paper/2508.13073/) (2025) - [FLOWER: Democratizing Generalist Robot Policies with Efficient Vision-Language-Action Flow Policies](https://mlsanity.com/paper/2509.04996/) (2025) - [Diffusion Transformer Policy](https://mlsanity.com/paper/2410.15959/) (2024) - [Diffusion Models for Robotic Manipulation: A Survey](https://mlsanity.com/paper/2504.08438/) (2025) - [Much Ado About Noising: Dispelling the Myths of Generative Robotic Control](https://mlsanity.com/paper/2512.01809/) (2025) - [CronusVLA: Towards Efficient and Robust Manipulation via Multi-Frame Vision-Language-Action Modeling](https://mlsanity.com/paper/2506.19816/) (2025) ## Built on by (most cited first, 4 of 4) - [ST-WAM: Semantic-Temporal World Action Model for Robust Manipulation under Visual Distribution Shifts](https://mlsanity.com/paper/2607.28993/) (2026) - [Frequency-Conditioned Flow Matching for Vision-Language-Action Models](https://mlsanity.com/paper/2609.10405/) (2026) - [RAVEL: Asynchronous Rolling Inference for Flow-Based Vision-Language-Action Models](https://mlsanity.com/paper/2609.34170/) (2026) - [Discrete Forcing: Infusing Discrete Guidance into Continuous Denoising for Few-Step Action Experts](https://mlsanity.com/paper/2609.39526/) (2026) ## Abstract (arXiv) Flow-based vision-language-action (VLA) policies offer strong expressivity for action generation, but suffer from a fundamental inefficiency: multi-step inference is required to recover action structure from uninformative Gaussian noise, leading to a poor efficiency-quality trade-off under real-time constraints. We address this issue by rethinking the role of the starting point in generative action modeling. Instead of shortening the sampling trajectory, we propose CF-VLA, a coarse-to-fine two-stage formulation that restructures action generation into a coarse initialization step that constructs an action-aware starting point, followed by a single-step local refinement that corrects residual errors. Concretely, the coarse stage learns a conditional posterior over endpoint velocity to transform Gaussian noise into a structured initialization, while the fine stage performs a fixed-time refinement from this initialization. To stabilize training, we introduce a stepwise strategy that first learns a controlled coarse predictor and then performs joint optimization. Experiments on CALVIN and LIBERO show that our method establishes a strong efficiency-performance frontier under low-NFE (Number of Function Evaluations) regimes: it consistently outperforms existing NFE=2 methods, matches or surpasses the NFE=10 π0.5 baseline on several metrics, reduces action sampling latency by 75.4%, and achieves the best average real-robot success rate of 83.0%, outperforming MIP by 19.5 points and π0.5 by 4.0 points. These results suggest that structured, coarse-to-fine generation enables both strong performance and efficient inference. Our code is available at https://github.com/EmbodiedAI-RoboTron/CF-VLA.