VLMLight: Safety-Critical Traffic Signal Control via Vision-Language Meta-Control and Dual-Branch Reasoning Architecture

Maonan Wang The Chinese University of Hong Kong, Shenzhen, ChinaShanghai AI Laboratory, Shanghai, China, Yirong Chen Shanghai AI Laboratory, Shanghai, China, Aoyu PangThe Chinese University of Hong Kong, Shenzhen, China, Yuxin CaiNanyang Technological University, Singapore, Chung Shue ChenNokia Bell · 2025

Read VLMLight: Safety-Critical Traffic Signal Control via Vision-Language Meta-Control and Dual-Branch Reasoning Architecture by Maonan Wang The Chinese University of Hong Kong, Shenzhen, ChinaShanghai AI Laboratory, Shanghai, China, Yirong Chen Shanghai AI Laboratory, Shanghai, China, Aoyu PangThe Chinese University of Hong Kong, Shenzhen, China, Yuxin CaiNanyang Technological University, Singapore, Chung Shue ChenNokia Bell (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-19486.txt.

Description

Maonan Wang and co-authors present a safety-first traffic light controller in this 2025 study. City signals must cut delays without endangering anyone, and learning-based controllers can falter in rare, risky scenes. Their system, VLMLight, watches intersections through a new image-based simulator that shows several views at once, so decisions can use vehicle type, motion, and crowding.

A large language model acts as meta-controller, judging each moment. Calm traffic goes to a fast learning policy. Hard cases switch to a reasoning branch where several model agents check phases, clear paths for emergency vehicles, and verify the rules. In tests, waits for emergency vehicles fell by up to sixty-five percent versus learning-only control. Normal traffic lost almost nothing, keeping the system real-time capable.

Catalog record

CREATORMaonan Wang The Chinese University of Hong Kong, Shenzhen, ChinaShanghai AI Laboratory, Shanghai, China, Yirong Chen Shanghai AI Laboratory, Shanghai, China, Aoyu PangThe Chinese University of Hong Kong, Shenzhen, China, Yuxin CaiNanyang Technological University, Singapore, Chung Shue ChenNokia Bell
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE62 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-19486

Browse the full library · About the Tower of Babel