Reflex: Fast Robot Policy Serving by Reusing and Correcting Past Actions
cs.RO, cs.AI
Submitted: 2026-10-31
Updated: 2026-10-31
Code: https://github.com/TheRobotStudio/SO-ARM100
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Efficient Vision-Language-Action Management and Serving for Robot Factories
- Action Chunk Scheduling for Batched Robot Policy Serving
- SoundWeaver: Compositional Warm-Starting for Text-to-Audio Diffusion Serving
- $\pi_\texttt{RL}$: Online RL Fine-tuning for Flow-based Vision-Language-Action Models
- Kairos: A Scalable Serving System for Physical AI
- LIBERO-Plus: In-depth Robustness Analysis of Vision-Language-Action Models
- Unified 4D World Action Modeling from Video Priors with Asynchronous Denoising
- AdaVLA: Adaptive Step Flow Matching for Training-free Acceleration of Vision-Language-Action Models
- VLA-Rolling: Fewer Denoising Steps, Better Control via Adaptive Action Reuse
- ROSA: A Robotics Foundation Model Serving System for Robot Factories
- Efficient World Action Model Inference with Adaptive Intermediate States
- GR00T N1: An Open Foundation Model for Generalist Humanoid Robots
- ActionCache: Training-Free Acceleration for Vision-Language-Action Models with Action Caching and Refinement
- SmolVLA: A Vision-Language-Action Model for Affordable and Efficient Robotics
- Fast-WAM: Do World Action Models Need Test-time Future Imagination?
- Retrieve-then-Steer: Online Success Memory for Test-Time Adaptation of Generative VLAs
- Leveraging OS-Level Primitives for Robotic Action Management
Related papers
- Rethinking Actions: Continuity-Aware Priors for World-Model Planning
- MOSAIC: Modular Obstacle and Spatial-Aware Framework for Manipulation In Cluttered Scenes
- SAGAS: Semantic-Aware Graph-Assisted Stitching for Offline Temporal Logic Planning
- DexSeed: Expanding Human Hand-Object Interactions into Robot Demonstrations for Dexterous Policy Learning
- LaST0.5: A Generalist World Action Model with Interaction-Centric Latent Reasoning
- CIALLO: Learning Shared Intent Representations for Cross-Embodiment Manipulation