ITSC 2025 Paper Abstract

Paper TH-LA-T22.3

Im, Jaegyun (soonchunhyang university), Kim, Byeonghun (Soonchunhyang University), Jin, Joobin (Soonchunhyang university), Noh, Byeongjoon (Soonchunhyang University)

Traffic Context-Augmented Vehicle Trajectory Prediction Framework Using Multimodal LLM

Scheduled for presentation during the Invited Session "S22c-Emerging Trends in AV Research" (TH-LA-T22), Thursday, November 20, 2025, 16:40−17:00, Coolangata 1

2025 IEEE 28th International Conference on Intelligent Transportation Systems (ITSC), November 18-21, 2025, Gold Coast, Australia

This information is tentative and subject to change. Compiled on October 18, 2025

Keywords AI, Machine Learning for Dynamic Traffic Signal Control and Optimization, Deep Learning for Scene Understanding and Semantic Segmentation in Autonomous Vehicles, Real-time Object Detection and Tracking for Dynamic Traffic Environments

Abstract

We propose a novel multimodal framework for vehicle trajectory prediction that integrates drone imagery and user-defined textual prompts through a large language model (LLM). The framework comprises three key modules: (1) Trajectory Information Extraction (TIE), which detects and encodes vehicle movements; (2) Traffic Scene Understanding Extraction (TSUE), which combines visual and textual context; and (3) Multimodal Fusion Prediction (MFP), which produces context-aware predictions. Experiments conducted on a high-resolution drone-view dataset show that the proposed method outperforms existing approaches, while ablation studies highlight the importance of multimodal representation and parameter-efficient fine-tuning using LoRA in modeling complex traffic dynamics.