Efficient VLA Inference

VLA: Vision-Language-Action Model

Overview

This topic studies how to run vision-language-action models fast enough for responsive, reliable robot control, including on edge devices. Common approaches prune or cache visual tokens, quantize or distill models, reduce action-generation steps, and overlap inference with observation and execution. Methods are tested on manipulation benchmarks such as LIBERO and RoboTwin, as well as real robots, using measures like latency, control frequency, memory, energy, and task success. The main tension is that faster inference can undermine closed-loop reliability or motion quality, so gains must be assessed in full task execution rather than by model speed alone.

Latest papers

CardsList

221 papers36 papers in the last four weeks, down 35% on the four weeks before. 0.3% of all new papers.

  1. VSpark: Speculative Decoding for Autoregressive Vision-Language-Action Models

    Nov 17, 2026Suho Yoo, Youngjoon Jang, Hyebin Cho +3Efficient VLA InferenceVision-Language-Action Models

  2. SR-VLA: Stall-Reduced VLA Execution via Predicted-Input Lookahead and Selective KV Cache Repair

    Nov 17, 2026Tao Zhang, Yi Zhong, Jinhao Ruan +4Efficient VLA InferenceLanguage Model Inference Caching

  3. POKE: Probe Offline, Keep Efficiently for Vision-Language-Action Models

    Nov 16, 2026Jaehyuk Lee, Yeajin Lee, Suhyun Bae +1Efficient VLA InferenceVisual Token Pruning

  4. Stochastic Self-Distillation for Flow-Based Vision-Language-Action Models

    Nov 16, 2026Jingru Guo, Alessandra Russo, Joey Bose +1Flow-Based VLAsEfficient VLA Inference

  5. SuperGlance: Driving at a Glance with a Vision-Language Model

    Nov 14, 2026Jieyuan Pei, Xin Yan, Mingwei Xu +9VLMs for Autonomous DrivingEfficient VLA Inference

  6. Wireless Edge-Assisted Adaptive Latent Reasoning for Vision-Language-Action Models in Physical AI

    Nov 14, 2026Fatih Temiz, Melike Erol-KantarciEfficient VLA InferenceReasoning for VLAs

  7. Vision-Language-Action Models: A Life-Cycle Survey of Design, Training and Deployment

    Nov 13, 2026Suvarna Kadam, Soran ParsaVision-Language-Action ModelsVLA Training

  8. Learning Low-Latency Action Verification for Frozen Vision-Language-Action Models

    Nov 10, 2026Wen-Han Hsieh, Ken Jiankun Zheng, Chia-Chien Chen +2Robotic ManipulationVision-Language-Action Models

  9. APS-VLA: Action-Guided and Future-Aware Token Pruning for Efficient Vision-Language-Action Models

    Nov 9, 2026Xiaoyang Xing, Chun-Mei Feng, Anthony Cohn +2Efficient VLA InferenceVisual Token Pruning

  10. Decoupled Asynchronous System for Humanoids (DASH): Latency-Agnostic Scaling of Vision-Language-Action Models for Humanoid Manipulation

    Nov 9, 2026Bingzhang Wang, Nathan Zhao, Jialeng Ni +4Closed-Loop Control of VLAsRobotic Manipulation

  11. Efficient VLA Inference via Retrieval-Based Speculation and Hardware Acceleration

    Nov 8, 2026Zhengyang Cheng, He Zhu, Zeyu Guo +6Efficient VLA InferenceAI Accelerator Inference

  12. Control-Aware Command Generation for Asynchronous Control and Action Inference with Vision-Based Policies towards Fast Execution

    Nov 8, 2026Takayuki Osa, Kazuki OtaRobotic ControlVision-Based Robot Control

  13. Where Should Models Run? A Collaborative Inference Framework for Embodied Intelligence in 6G

    Nov 7, 2026Wanlu Sun, Zhenhui Yuan, Xiang Ji +3Edge-Cloud Collaborative InferenceEfficient VLA Inference

  14. VLA Handover: Failure-Prediction-Based Switching from On-Robot to Server VLA Models

    Nov 6, 2026Takayuki NishioEfficient VLA InferenceFailure Detection for VLAs

  15. ActPrune: Action-Grounded Visual Token Pruning for Efficient Vision-Language-Action Models

    Nov 6, 2026Sungwon Moon, Minji Kim, Hyemin Boo +2Efficient VLA InferenceVisual Token Pruning

  16. VFlash: Fast, Accurate Visual Planning via Flash Primitive Decoding

    Nov 5, 2026Yu-Cheng Chou, Shihao Wang, Jin Huang +16Efficient VLA InferenceRobot Motion Prediction

  17. T3R: Training-Free Two-Stage Token Refinement Towards Efficient and Robust VLA Models

    Nov 5, 2026Mohammed Hassan, Zhenyang Chen, Zheng Wang +5Efficient VLA InferenceVisual Token Pruning

  18. Where Is VLA Inference Time Spent? Bottlenecks and Optimization Interactions Across GPUs

    Nov 4, 2026Muhammad Danish Waseem, Ahmed Ali-Eldin HassanEfficient VLA InferenceInference-Time Optimization

  19. Asynchronous Robot Policies

    Nov 3, 2026Sombit Dey, Luc Van Gool, Danda Pani PaudelVLMs for RoboticsClosed-Loop Control of VLAs

  20. LoopQ: Fair, Closed-Loop Evaluation of Quantized Vision-Language-Action Models

    Nov 3, 2026Syed Umer Bukhari, Magnus Almgren, Ahmed Ali-EldinPost-Training QuantizationEfficient VLA Inference

  21. VLALagBench: Understanding When and Why VLA Inference Latency Matters

    Nov 2, 2026Qi Wen, Borui Xu, Hussein Amro +11VLMs for RoboticsEfficient VLA Inference

  22. Scene-Adaptive Readout and Decoding in Diffusion Driving VLAs

    Nov 1, 2026Seungjun Yu, Jiho Choi, Inha Kang +3Diffusion-Based VLAsVision-Language-Action Models

  23. Vision-Language Models Are Joint Denoisers in Robotic Action Generation

    Oct 31, 2026Xicheng Gong, Wenxuan Song, Chengze Liu +5Diffusion-Based VLAsVision-Language-Action Models

  24. Leap-VLA: High-Rate Robot Control from a Low-Rate VLM Backbone

    Oct 31, 2026Lin Yang, Yunjie Shen, Zheyu Li +8Closed-Loop Control of VLAsEfficient VLA Inference

  25. STEER: Route-Aware Adaptive Reasoning for Autonomous Driving

    Oct 31, 2026Yangang Zou, Pei Liu, Nan Song +6Reasoning for VLAsEfficient VLA Inference

  26. GUI-JEPA: Candidate-Conditioned JEPA Latents as a Fast Action Policy for Mobile GUI Agents

    Oct 27, 2026Bingchao Wang, Yongcheng Jing, Shunyu Liu +5Mobile GUI AgentsMobile GUI Automation