Semantic Communications and Multimodal LLMs for Intelligent Traffic Monitoring

Traffic monitoring is essential for connected vehicle systems, enabling safer mobility, reduced congestion, and adaptive decision-making. This talk presents a semantic communication framework that integrates multimodal large language models (LLMs) with the digital twin of an autonomous vehicle test track. By focusing on meaning rather than raw data, the system transmits compact, context-rich representations to the cloud, where visual and language understanding enables accurate detection of vehicles, pedestrians, and traffic conditions. This approach demonstrates how semantic communication and AI-driven reasoning can work together to achieve efficient, scalable, and intelligent traffic monitoring in next-generation transportation networks.