Toward Real-Time VLAs: Stage-Aware Two-Step Flow Denoising and System-Level Evaluation

arXiv:2609.39822 · cs.RO · Submitted 2026-09-30 · Read on arXiv

cs.RO

Submitted: 2026-09-30

Updated: 2026-10-03

Code: https://github.com/MagiclabRobotics/Inference

Project page: https://embodied.magiclab.top/works/inference/index.html

Terminology

• Flow Matching — 39× in this paper · explained in 8 episodes
• Alignment — 13× in this paper · explained in 1 episode
• Action Chunking — 12× in this paper · explained in 1 episode
• Consistency — 8× in this paper · explained in 3 episodes
• Distillation — 7× in this paper · explained in 2 episodes
• Calibration — 6× in this paper · explained in 7 episodes
• Provenance — 3× in this paper · explained in 4 episodes
• Tokenization — 3× in this paper · explained in 2 episodes
• Verification — 3× in this paper · explained in 2 episodes
• World Model — 3× in this paper · explained in 7 episodes
• Accuracy — 2× in this paper · explained in 1 episode
• Action Flow — 2× in this paper · explained in 1 episode
• Consistency Distillation — 2× in this paper · explained in 2 episodes
• Generalization — 2× in this paper · explained in 13 episodes
• Self-Distillation — 2× in this paper · explained in 7 episodes
• Action Space — 1× in this paper · explained in 1 episode
• Action-Conditioned World Model — 1× in this paper · explained in 1 episode
• Generative Modeling — 1× in this paper · explained in 3 episodes
• Phase Transitions — 1× in this paper · explained in 2 episodes
• Quantization — 1× in this paper · explained in 12 episodes
• Regularization — 1× in this paper · explained in 3 episodes
• Robustness — 1× in this paper · explained in 5 episodes
• Tool Center Point — 1× in this paper
• Two-Stage Framework — 1× in this paper · explained in 2 episodes
• Unified Framework — 1× in this paper · explained in 5 episodes
• Unified Robot Description Format — 1× in this paper
• VLA Models — 1× in this paper · explained in 3 episodes
• Vision-Language-Action (VLA) Models — 1× in this paper · explained in 2 episodes

Sources

• Understanding Asynchronous Inference Methods for Vision-Language-Action Models
• Training-Time Action Conditioning for Efficient Real-Time Chunking
• Riemannian Flow Matching Policy for Robot Motion Learning
• ImplicitRDP: An End-to-End Visual-Force Diffusion Policy with Structural Slow-Fast Learning
• DREAM-Chunk: Reactive Action Chunking with Latent World Model
• Where Should Action Generation Begin? A Learnable Source Prior for Generative Robot Policies
• Fast and Robust Visuomotor Riemannian Flow Matching Policy
• Denoising Tells When to Replan: Denoising-Variance Adaptive Chunking for Flow-Based Robot Policies
• Reflex: Real-Time VLA Control through Streaming Inference
• B-spline Policy: Accelerating Manipulation Policies via B-spline Action Representations
• Action-to-Action Flow Matching
• FutureRTC: Real-Time Robot Execution with Anticipatory-Conditioned Action Chunking
• AsyncVLA: Asynchronous Flow Matching for Vision-Language-Action Models
• Mixture of Horizons in Action Chunking
• WarmPrior: Straightening Flow-Matching Policies with Temporal Priors
• Flow Matching for Generative Modeling
• Bidirectional Decoding: Improving Action Chunking via Guided Test-Time Sampling
• CheckVLA: Execution-Time Verification with Action-Conditioned World Model for Long-Horizon Mobile Manipulation
• FASTER: Rethinking Real-Time Flow VLAs
• SnapFlow: One-Step Action Generation for Flow-Matching VLAs via Progressive Self-Distillation

Related papers

• FMT x: An Efficient and Asymptotically Optimal Extension of the Fast Marching Tree for Dynamic Replanning
• MPCFormer: A physics-informed data-driven approach for explainable socially-aware autonomous driving
• RoboLab: A High-Fidelity Simulation Benchmark for Analysis of Task Generalist Policies
• HRDexDB: A 4D Dexterous Grasping Dataset Across Human and Multiple Robot Embodiments
• APT: Action Expert Pretraining Improves Instruction Generalization of Vision-Language-Action Policies
• Fine-tuning is Not Enough: A Parallel Framework for Collaborative Imitation and Reinforcement Learning in End-to-end Autonomous Driving