Diffusion-Based VLA Models

VLA: Vision-Language-Action

Follow this topic Ask about this topic

Latest papers

8 papers2 papers in the last four weeks, against 1 the four weeks before. 0.0% of all new papers.

Scene-Adaptive Readout and Decoding in Diffusion Driving VLAs
Nov 1, 2026Seungjun Yu, Jiho Choi, Inha Kang +3Diffusion-Based VLA Models Vision-Language-Action Models
Korea Advanced Institute of Science and Technology (KAIST)
Vision-Language Models Are Joint Denoisers in Robotic Action Generation
Oct 31, 2026Xicheng Gong, Wenxuan Song, Chengze Liu +5Diffusion-Based VLA Models Vision-Language-Action Models
Peking University · The Hong Kong University of Science and Technology (Guangzhou) · Microsoft Research Asia
Humanoid Loco-Manipulation With Discrete VLA Model
Sep 28, 2026Wenxin Shao, Siqi Chai, Kun Li +4Humanoid Loco-Manipulation Vision-Language-Action Models
Horizon Robotics
GUARD: Grounding Uncertainty and Ablation-Based Risk Detection for Diffusion-Based VLAs
Aug 5, 2026Suhas Hegde, Jitendra Yasaswi Bharadwaj KattaFailure Detection for VLA Models Uncertainty Estimation for VLA Models
Bosch Research India
dVLA-RL: Reinforcement Learning over Denoising Trajectories for Discrete Diffusion Vision-Language-Action Models
Jun 22, 2026Yuhao Wu, Yitian Liu, Weijie Shen +13Vision-Language-Action Models Diffusion-Based VLA Models
ScaleLab at SJTU · 1Shanghai Jiao Tong University · 3Baidu AI Cloud +3
Realtime-VLA FLASH: Speculative Inference Framework for Diffusion-based VLAs
May 13, 2026Jiahui Niu, Kefan Gu, Yucheng Zhao +5Efficient VLA Model Inference Speculative Decoding
State Key Lab of Processors, Institute of Computing Technology, CAS · University of Chinese Academy of Sciences · Dexmal +1
MMaDA-VLA: Large Diffusion Vision-Language-Action Model with Unified Multi-Modal Instruction and Generation
Mar 26, 2026Yang Liu, Pengxiang Ding, Tengyue Jiang +10Diffusion-Based VLA Models Vision-Language-Action Models
Westlake University Hangzhou, China · Zhejiang University Hangzhou, China · East China University of Science and Technology Shanghai, China +2
TokenPB: Enabling VLA Action Self-Verification via VLM Token Probability Bucketing
Mar 18, 2026Chen Zhao, Zhuoran Wang, Haoyang Li +6Diffusion-Based VLA Models Vision-Language-Action Models
School of Information, Renmin University of China · Key Laboratory of Data Engineering and Knowledge Engineering, Beijing, China · Beijing University of Posts and Telecommunications +1

Nov 1, 2026J/K move · Enter open · S save

Scene-Adaptive Readout and Decoding in Diffusion Driving VLAs

Seungjun Yu, Jiho Choi, Inha Kang, Hojun Choi+2

Korea Advanced Institute of Science and Technology (KAIST)

Diffusion-Based VLA Models Vision-Language-Action Models VLMs for Autonomous Driving Efficient VLA Model Inference Autonomous Driving Planning Adaptive Computation for Diffusion Models

Open paper AskPDF arXiv

Diffusion driving vision-language-action models repeatedly execute the backbone at every denoising stage, incurring substantial inference cost, while it remains unclear whether all scenes require the same decoding computation. To distinguish when a trajectory becomes readable from how much further refinement it requires, we analyze trajectory readout across network depth and decoding progress. We identify a depth-decoding dissociation in which late-layer representations already support useful trajectory readout after the first forward sweep, whereas early-layer representations benefit more from continued decoding. We further find that both the preferred readout layer and decoding depth vary across scenes. Based on these observations, we propose a scene-adaptive readout and decoding framework. A scene-conditioned readout MoE adaptively weights and combines trajectory predictions from layer-specific late-layer representations after a single backbone sweep, while an adaptive decoding policy applies additional native refinement only when required by the scene. We evaluate our framework on two diffusion driving VLAs across the Waymo E2E and nuScenes benchmarks. Our method improves validation RFS by up to 0.56 while also improving test performance, and achieves an end-to-end speedup of up to

6.0\times

on Waymo E2E, while consistently reducing trajectory error on nuScenes. These results demonstrate that scene-level readout and decoding allocation improves planning quality while avoiding unnecessary decoding. Our code is available at https://github.com/sjyu001/SARD