The principles and implementation of automotive digital twin in autonomous driving systems

In autonomous driving systems, the integration of advanced technologies such as artificial intelligence, sensor fusion, and advanced control systems is rapidly advancing. Digital Twin (DT) technology is reshaping the automotive field by integrating real-time sensing, distributed intelligence, and human-centric decision-making processes. Despite this progress, real-time, efficient, and safe decision-making processes still pose significant challenges, and current research on Vehicle-Driver-Road Digital Twins (VDT, DrDT, and RDT) remains fragmented, limiting the potential to deliver coherent, safety-focused mobility solutions. This study presents a comprehensive synthesis of recently published researches, examining the definitions, architectures, and interaction paradigms of VDT, DrDT, and RDT, and aims to provide a framework that governs the decision-making process of an autonomous vehicle with a digital twin-based approach. Based on these findings, the conceptual foundations of a unified, safety-focused Automotive Digital Twin (ADT) framework are laid out, paving the way towards scalable, reliable, and human-centric digital ecosystems. This thesis aims to develop a Digital Twin with a systematic, conceptual, and architectural framework for providing real-time decision support for autonomous driving systems, enabling the monitoring, analysis, and control of vehicle behavior using Digital Twins, and providing runtime and safety-aware models for Automotive Digital Twins.