arXiv:2606. 24759v1 Announce Type: cross Abstract: Recent multimodal large language models (MLLMs) have shown strong potential for autonomous driving scene understanding, yet existing methods still face a fundamental trade-off between temporal reasoning and spatial precision.
By Xiaowei Gao, Pengxiang Li, Yitai Cheng, Ruihan Xu, James Haworth, Stephen Law, Yun Ye
Recent multimodal large language models (MLLMs) have shown strong potential for autonomous driving scene understanding, yet existing methods still face a fundamental trade-off between temporal reasoning and spatial precision. Models that rely on single-frame or low-resolution inputs often miss small, distant, or partially occluded hazards, while language-centric driving models frequently provide limited grounded evidence for their explanations.
M3-Former is a multimodal transformer framework that uses large language models to encode vessel static attributes and navigational intent as semantic priors for long‑term trajectory prediction. It builds a unified multimodal representation space, aligns static semantic information with dynamic trajectory features via self‑attention, and employs a dual‑granularity Mixture‑of‑Experts architecture to capture both global route planning and fine‑grained maneuvering behaviors. A Steering‑Weighted Cross‑Entropy loss further improves accuracy on sparse turning events, and experiments on a Danish AIS dataset show consistent improvements over state‑of‑the‑art baselines, reducing ADE and FDE by up to 5.1% in 4‑hour predictions.
By Wenzhe Jin, Haina Tang
The paper introduces FloodVision, a knowledge-guided vision‑language framework that estimates urban flood depth from a single RGB image. It combines a general‑purpose vision‑language model with FloodKG, a domain knowledge base that encodes canonical object dimensions and component landmarks to promote component‑level reasoning. On 654 crowdsourced New York flood images, FloodVision cuts mean absolute error from 15.62 cm to 8.75 cm and median error from 14.35 cm to 7.75 cm, outperforming the VLM‑only baseline in 69.3 % of cases.
By Zhangding Liu, Neda Mohammadi, John E. Taylor
arXiv:2607. 23537v1 Announce Type: new Abstract: Autonomous driving under adverse weather remains a critical challenge, yet existing vision-language benchmarks mainly evaluate under standard conditions, synthetic corruptions, or single modality.
By Qiao Yan, Yihan Wang, Zhenghao Xing, Jiaqi Xu, Pheng-Ann Heng
The paper investigates why multimodal large language models (MLLMs) struggle with vision‑centric tasks when visual evidence conflicts with pretrained language knowledge. Using image reconstruction and a new WhatIfVis benchmark, the authors show that MLLMs preserve coarse‑grained visual attributes but fail to consistently use them, and that supervised fine‑tuning and activation patching can improve controllability of visual context sensitivity. The study demonstrates that the main bottleneck lies in the models’ inability to reliably regulate their reliance on visual evidence rather than in visual perception itself.
By Jiaang Li, Chengzu Li, Zhaochong An, Yifei Yuan, Xi Liu, Serge Belongie, V\'esteinn Sn{\ae}bjarnarson