ITSC 2025 Paper Abstract

Paper TH-EA-T23.5

Gong, Junjie (Southeast University), Zhang, Qixiang (Southeast University), liang, jinhao (Southeast University), Yao, Mingxi (Southeast University), Wang, Jinxiang (Southeast University), Yin, Guodong (Southeast University)

Personalized Takeover Request Strategies for Level 3 Autonomous Vehicles Considering Driver Style Differences

Scheduled for presentation during the Invited Session "S23b-Trustworthy AI for Traffic Sensing and Control" (TH-EA-T23), Thursday, November 20, 2025, 14:50−14:50, Coolangata 2

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 Human-Machine Interaction Systems for Enhanced Driver Assistance and Safety, Driver Behavior Monitoring and Feedback Systems for Semi-autonomous Vehicles

Abstract

Level 3 autonomous driving (AD) represents a feasible implementation of intelligent vehicles, requiring prompt driver intervention when the operational design domain is exceeded. Effective Takeover Requests (TORs) are essential to ensure safe control handover and prevent potential risks. However, most strategies do not fully consider individual driving styles, which may increase driving risks. To this end, this paper proposes a semi-supervised Gaussian mixture model to classify driving styles, integrating subjective and objective data. A 3 × 2 × 3 lane-change avoidance experiment is designed to investigate the impact of different TOR on driver performance, with warning modality and content as independent variables. Results show that conservative drivers perform optimally with auditory decision-making cues and no visual content. In contrast, aggressive drivers display more cautious behaviors with a combination of auditory and visual decision-making cues, despite a preference for simpler prompts. Moderate drivers exhibit the fastest response times and highest satisfaction when auditory decision-making cues were provided without visual content. These findings underscore the necessity of tailoring TOR strategies to individual driving styles, thereby improving the handover safety and driving performance in Level 3 AD systems.