ITSC 2025 Paper Abstract

Paper TH-EA-T29.1

Lao, Xinyun (Tongji University), Gao, Jincheng (Tongji University), Shen, Yu (Tongji University), Ding, Chenhang (Tongji University), Ji, Yuxiong (Tongji University)

Driver-Vehicle Visual Fusion Enabling Safer Conflict Avoidance at Unsignalized T-Intersections

Scheduled for presentation during the Regular Session "S29b-Human Factors and Human Machine Interaction in Automated Driving" (TH-EA-T29), Thursday, November 20, 2025, 13:30−13:50, Currumbin

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 Advanced Sensor Fusion for Robust Autonomous Vehicle Perception, Human-Machine Interaction Systems for Enhanced Driver Assistance and Safety

Abstract

While autonomous vehicles (AVs) aim to perceive as many environmental objects as possible, drivers instinctively prioritize safety-critical elements. This study utilizes the cognitive advantage of drivers by integrating human attention into the AV perception system to enhance driving performance, especially in complex traffic scenarios. We propose a three-stage fusion framework that systematically embeds driver gaze patterns into AV vision systems. By utilizing driver gaze point tracking data, the system effectively prioritizes critical risk elements like oncoming vehicles, thereby enhancing the vehicle's visual sensing capabilities. To validate our approach, we conducted a series of left-turn field tests at an unsignalized T-intersection, involving 19 licensed drivers operating a Level 2 AV for data collection and visual fusion. The safety impacts of the fusion model were subsequently evaluated using a virtual simulation platform, where the AV's longitudinal movements were controlled based on the integrated risk field perceived by both the driver and the vehicle's sensors. Conflict risks were analyzed by comparing Time-To-Collision (TTC) indicators pre- and post-fusion. Results show that the TTC increased from 2.49 s to 5.74 s, indicating that the proposed visual perception fusion model effectively reduces potential vehicle conflict risks under identical traffic conditions and enhances the safety of autonomous driving.