Efficient VLA Model Inference

VLA: Vision-Language-Action

Ask about this topic

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

190 papers56 papers in the last four weeks, up 115% on the four weeks before. 0.3% of all new papers.

  1. VLA-InfoEntropy: A Training-Free Vision-Attention Information Entropy Approach for Vision-Language-Action Models Inference Acceleration and Success

    Apr 7, 2026Chuhang Liu, Yayun He, Zuheng Kang +2Efficient VLA Model InferenceVision-Language-Action Models

    Ping An Technology (Shenzhen) Co., Ltd., Shenzhen, China · Tsinghua Shenzhen International Graduate School, Tsinghua University, Shenzhen, China

  2. Open-Loop Planning, Closed-Loop Verification: Speculative Verification for VLA

    Apr 3, 2026Zihua Wang, Zhitao Lin, Ruibo Li +4Closed-Loop Control of VLA ModelsVision-Language-Action Models

    School of Computer Science and Engineering, Southeast University, Nanjing, China · Nanyang Technological University, Singapore · School of Cyber Science and Engineering, Southeast University, Nanjing, China +1

  3. MobiGuide: Vocabulary-Guided Online Recovery for Mobile Manipulation

    Apr 2, 2026Jia Syuen Lim, Zhizhen Zhang, Peter Bohm +3Mobile ManipulationDiffusion Policy

    UQMM Lab, The University of Queensland, Brisbane, Australia · Robotics and Autonomous Systems Group, CSIRO, Brisbane, Australia

  4. StreamingVLA: Streaming Vision-Language-Action Model with Action Flow Matching and Adaptive Early Observation

    Mar 30, 2026Yiran Shi, Dongqi Guo, Tianchen Zhao +8Vision-Language-Action ModelsEfficient VLA Model Inference

    Tsinghua University · Lenovo Group Ltd.

  5. Realtime-VLA V2: Learning to Run VLAs Fast, Smooth, and Accurate

    Mar 27, 2026Chen Yang, Yucheng Hu, Yunchao Ma +3Robotic ControlVLMs for Robotics

    Dexmal

  6. Beyond Attention Magnitude: Leveraging Inter-layer Rank Consistency for Efficient Vision-Language-Action Models

    Mar 26, 2026Peiju Liu, Jinming Liu, Xipeng Qiu +1Efficient VLA Model InferenceVision-Language-Action Models

    Fudan University

  7. Training-Free Interaction-Aligned Visual Token Pruning for Efficient Embodied Manipulation

    Mar 24, 2026Jintao Cheng, Weibin Li, Haozhe Wang +7Visual Token PruningEfficient VLA Model Inference

    Department of Electronic and Computer Engineering, The Hong Kong University of Science and Technology, Hong Kong, China · University of Macau, Macau, China · School of Data Science and Engineering, Xingzhi College, South China Normal University, Guangzhou, China +3

  8. Agile-VLA: Few-Shot Industrial Pose Rectification via Implicit Affordance Anchoring

    Mar 24, 2026Teng Yan, Zhengyang Pei, Chengyu Shi +8Affordance-Based Robotic ManipulationRobotic Manipulation

    The Hong Kong University of Science and Technology (Guangzhou)

  9. DualCoT-VLA: Visual-Linguistic Chain of Thought via Parallel Reasoning for Vision-Language-Action Models

    Mar 23, 2026Zhide Zhong, Junfeng Li, Junjie He +10VLMs for RoboticsReasoning for VLA Models

    The Hong Kong University of Science and Technology (Guangzhou) · Huawei Foundation Model Department

  10. RoboECC: Multi-Factor-Aware Edge-Cloud Collaborative Deployment for VLA Models

    Mar 21, 2026Zihao Zheng, Hangyu Cao, Jiayu Chen +6Edge-Cloud Collaborative InferenceEfficient VLA Model Inference

    School of Computer Science, Peking University, Beijing, China · School of Computer Science, South China University of Technology, Guangzhou, China · School of Artificial Intelligence, Beijing Normal University, Beijing, China +1

  11. FASTER: Rethinking Real-Time Flow VLAs

    Mar 19, 2026Yuxiang Lu, Zhe Liu, Xianzhe Fan +5Efficient VLA Model InferenceRobotics

    The University of Hong Kong · ACE Robotics

  12. From Inference Efficiency to Embodied Efficiency: Revisiting Efficiency Metrics for Vision-Language-Action Models

    Mar 19, 2026Zhuofan Li, Hongkun Yang, Zhenyang Chen +4Efficient VLA ModelsVision-Language-Action Models

    Celine · Hong Kong University of Science and Technology · Georgia Institute of Technology

  13. ProbeFlow: Training-Free Adaptive Flow Matching for Vision-Language-Action Models

    Mar 18, 2026Zhou Fang, Jiaqi Wang, Wenzhuo Li +2Efficient VLA Model InferenceVision-Language-Action Models

    School of Computer Science and Engineering, Southeast University, China · Key Laboratory of New Generation Artificial Intelligence Technology and Its Interdisciplinary Applications, Ministry of Education, China · School of Electronic Science & Engineering, Southeast University, China

  14. HeiSD: Hybrid Speculative Decoding for Embodied Vision-Language-Action Models with Kinematic Awareness

    Mar 18, 2026Zihao Zheng, Zhihao Mao, Sicheng Tian +8Efficient VLA Model InferenceSpeculative Decoding

    School of Computer Science, Peking University · School of Computer Science, China University of Geosciences (Wuhan) · School of Artificial Intelligence, Beijing Normal University +1

  15. Embodied Foundation Models at the Edge: A Survey of Deployment Constraints and Mitigation Strategies

    Mar 16, 2026Utkarsh Grover, Ravi Ranjan, Mingyang Mao +9Robot Foundation ModelsEdge Computing

    University of South Florida, USA · Florida International University, USA · Johns Hopkins University, USA +2

  16. AnolisVLA: Lightweight Vision-Language-Action Model with Deep State Space Models for Mobile Manipulation

    Mar 16, 2026Yusuke Takagi, Motonari Kambara, Daichi Yashima +3Language-Conditioned Robot ManipulationVision-Language-Action Models

    Keio University, 3-14-1 Hiyoshi, Kohoku, Yokohama, Kanagawa 223-8522, Japan.

  17. AutoMoT: A Unified Vision-Language-Action Model with Asynchronous Mixture-of-Transformers for End-to-End Autonomous Driving

    Mar 16, 2026Wenhui Huang, Songyan Zhang, Qihang Huang +6Mixture-of-TransformersEnd-to-End Autonomous Driving

    Nanyang Technological University, Singapore · Harvard University, US · Xiaomi EV, China

  18. OxyGen: Unified KV Cache Management for VLA Inference under Multi-Task Parallelism

    Mar 15, 2026Xiangyu Li, Huaizhi Tang, Xin Ding +3Efficient VLA Model InferenceKV-Cache Management

    Institute for AI Industry Research (AIR), Tsinghua University · Department of Electronic Engineering, Tsinghua University · University of Science and Technology of China

  19. DynVLA: Learning World Dynamics for Action Reasoning in Autonomous Driving

    Mar 11, 2026Shuyao Shang, Bing Zhan, Yunfei Yan +9Autonomous DrivingVision-Language-Action Models

    NLPR, Institute of Automation, Chinese Academy of Sciences (CASIA) · Yinwang Intelligent Technology Co. Ltd.

  20. DepthCache: Depth-Guided Training-Free Visual Token Merging for Vision-Language-Action Model Inference

    Mar 11, 2026Yuquan Li, Lianjie Ma, Han Ding +1Efficient VLA Model InferenceVisual Token Compression

    School of Artificial Intelligence and Automation, Huazhong University of Science and Technology, Wuhan, China. · School of Mechanical Science and Engineering, Huazhong University of Science and Technology, Wuhan, China.

  21. ActECC: Action Chunk Temporal Harnessing for Edge-Cloud Collaborative VLA Inference System

    Mar 9, 2026Zihao Zheng, Zhihao Mao, Hui Huang +9Temporal Modeling for VLA ModelsEdge-Cloud Collaborative Inference

    School of Computer Science, Peking University, Beijing, China. · School of Artificial Intelligence, Beijing Normal University, Beijing, China. · School of Software Engineering, South China University of Technology, Guangzhou, China.

  22. DyQ-VLA: Temporal-Dynamic-Aware Quantization for Embodied Vision-Language-Action Models

    Mar 9, 2026Zihao Zheng, Hangyu Cao, Chibang Tao +9VLM QuantizationEfficient VLA Model Inference

    School of Computer Science, Peking University, Beijing, China. · School of Software Engineering, South China University of Technology, Guangzhou, China. · School of Artificial Intelligence, Beijing Normal University, Beijing, China. +1

  23. Critic in the Loop: A Tri-System VLA Framework for Robust Long-Horizon Manipulation

    Mar 5, 2026Pengfei Yi, Yingjie Ma, Wenjiang Xu +4Long-Horizon Robotic ManipulationVLMs for Robotics

    Institute of Automation, Chinese Academy of Sciences · the School of Artificial Intelligence, University of Chinese Academy of Sciences

  24. LiteVLA-Edge: Quantized On-Device Multimodal Control for Embedded Robotics

    Mar 3, 2026Justin Williams, Kishor Datta Gupta, Roy George +1Efficient VLA Model InferenceOn-Device Inference

    Clark Atlanta University, Atlanta, GA, USA · Siemens Corporation, Princeton, NJ, USA

  25. KERV: Kinematic-Rectified Speculative Decoding for Embodied VLA Models

    Mar 2, 2026Zihao Zheng, Zhihao Mao, Maoliang Li +6Robot KinematicsEfficient VLA Model Inference

    School of Computer Science, Peking University, Beijing, China · School of Computer Science, China University of Geosciences (Wuhan), Wuhan, China

  26. ATA: Bridging Implicit Reasoning with Attention-Guided and Action-Guided Inference for Vision-Language Action Models

    Mar 2, 2026Cheng Yang, Jianhao Jiao, Lingyi Huang +8Vision-Language-Action ModelsReasoning for VLA Models

    Rutgers University · University College London · Rice University +1

  27. Mean-Flow based One-Step Vision-Language-Action

    Mar 2, 2026Yang Chen, Xiaoguang Ma, Bin ZhaoEfficient VLA Model InferenceVision-Language-Action Models

  28. Unifying Language-Action Understanding and Generation for Autonomous Driving

    Mar 2, 2026Xinyang Wang, Qian Liu, Wenjie Ding +7Vision-Language-Action ModelsEnd-to-End Autonomous Driving

    State Key Lab of CAD&CG, Zhejiang University · Li Auto

  29. Characterizing VLA Models: Identifying the Action Generation Bottleneck for Edge AI Architectures

    Mar 1, 2026Manoj Vishwanathan, Suvinay Subramanian, Anand RaghunathanVision-Language-Action ModelsEfficient VLA Model Inference

    Google, Mountain View, CA USA · Purdue University, West Lafayette, IN USA

  30. DySL-VLA: Efficient Vision-Language-Action Model Inference via Dynamic-Static Layer-Skipping for Robot Manipulation

    Feb 26, 2026Zebin Yang, Yijiahao Qi, Tong Xie +3Layer Skipping in Language Model DecodingEfficient VLA Model Inference

    Institute for Artificial Intelligence, Peking University, Beijing, China · School of Integrated Circuits, Peking University, Beijing, China · School of Electronics Engineering and Computer Science, Peking University, Beijing, China +2