Postdoctoral Fellow
Research group | Networks and Distributed Systems
Main supervisor | Tor Skeie
Co-supervisor | -
Affiliation | Department of Informatics, UiO
Contact | zhihongc@ifi.uio.no
Short bio
I earned my Ph.D. from Shandong University (Jinan, China) between 2019 and 2024, where my research focused on dynamic recommendation systems, data mining, and large language models. During my Ph.D., I was a visiting scholar at the University of Technology Sydney (Sydney, Australia) from 2022 to 2023 under the supervision of Prof. Guandong Xu. Following that, I continued as a visiting scholar at Macquarie University (Sydney, Australia) from 2023 to 2024, working with Prof. Guanfeng Liu.
Currently, I am a Postdoctoral Fellow at the University of Oslo, where I work on the project Revolutionizing Autonomous Driving: An AI-Driven, Multifaceted Framework. The project aims to enhance the safety, adaptability, and efficiency of autonomous driving by equipping vehicles with learning and autonomous decision-making capabilities through AI and Digital Twin technologies.
Research interests and hobbies
My research interests include autonomous driving, digital twins, dynamic recommendation, large language models, and machine learning. I also enjoy reading books, drawing, and running.
DSTrain project
Revolutionizing Autonomous Driving: An AI-Driven, Multifaceted Framework
Autonomous Driving (AD) is revolutionizing the power and transport sectors, enabling vehicles to navigate independently using AI-driven decision-making. However, achieving high accuracy, safety, and adaptability remains a challenge. Digital Twin (DT) technology enhances AD through real-time simulation, environmental modeling, and predictive analytics, but persistent challenges hinder its effectiveness, including accurate perception of dynamic and complex environments, explainable interactions for improved human understanding, and data privacy assurance in multi-vehicle systems. Addressing these issues is crucial for the practical deployment of AD in real-world scenarios.
To overcome these challenges, I propose RADAR, an AI-driven, multifaceted framework that integrates DT, Large Language Models (LLMs), and federated learning. RADAR consists of three key components: DriveTwin, which enhances real-time digital twin modeling and improves generalization in AD; ExplainAuto, which utilizes LLMs and transparent decision-making to provide human-understandable explanations; and MultiAuto, which ensures secure multi-vehicle coordination with privacy-preserving federated learning and augmented reality guidance.
By bridging AI, DT, and explainability, RADAR pioneers a new interdisciplinary paradigm for AD, enhancing safety, adaptability, and human-AI interaction. It provides high-fidelity simulations, real-time decision transparency, and secure multi-agent collaboration, significantly improving AD’s reliability and real-world applicability. Furthermore, RADAR’s innovations in LLM-powered explainability and privacy-aware learning can extend beyond AD to smart cities, intelligent transportation, and AI-driven automation, fostering industry collaboration and market competitiveness.