CV

Ph.D. Student in Mechanical Engineering

• Mobility Systems Control (MSC) Laboratory, advised by Dr. Yunli Shao

B.E. in Computer Science and Engineering

• Highest Honors in Computer Science and Engineering; The Best Graduate Thesis Award
• Student Teaching Assistant for Object-Oriented Analysis and Design and Machine Learning (Most Outstanding TA, Fall 2023)

Optimizing Autonomous Driving Datasets for Detection with Smaller Data Sizes and Equivalent Effectiveness

• Zexuan Jia*, Gongjin Lan* (* equal contribution)

Joint Prediction of Virtual Leader from Future Occupancy and Queueing Theory in Automated Vehicles

• Zexuan Jia, Yunli Shao (preview version at ACC, IEEE Xplore)

LLM-based and Game-Theoretic Traffic Trajectory Generation for Autonomous Driving

• Zexuan Jia, Huayu Xu

Traffic-State Reconstruction and Prediction under Partial Observability using Hybrid AI Learning and Traffic Flow Models

• Develop a hybrid traffic-state reconstruction and prediction framework integrating deep learning with traffic flow modeling
• Design AI models to reconstruct short-term traffic occupancy and flow from partial observations in urban signalized environments
• Incorporate traffic flow dynamics to extend prediction accuracy to corridor-level horizons and capture underlying traffic evolution beyond observed data

A 3D Digital Twin Framework for Closed-Loop Simulation and Control of Autonomous Vehicles (Sponsored by DOE through ORNL)

• Develop a multi-resolution digital twin integrating SUMO and CARLA for closed-loop simulation of traffic, control, vehicle dynamics, and hardware-in-the-loop
• Propose a five-layer (L0–L4) fidelity framework to quantify the sim-to-real gap across geometry, sensor signals, perception, and behavior, with predictive-validity analysis and gap-budget attribution to trace error sources
• Build a ROS-free annotation toolkit for real-world multi-sensor logs (six cameras, 3D LiDAR) that aligns LiDAR-keyed frames and auto-seeds 3D bounding boxes, reducing manual labeling to box refinement

CDASim Upgrade to Support Cooperative Driving Automation Research (Sponsored by DOT through Leidos)

• Develop core components of a distributed co-simulation framework integrating CARLA–CARMA–SUMO–ROS2 for unified traffic, vehicle, sensing, and control stacks
• Conduct component-level and system-level testing to enable scenario-based validation of cooperative driving automation (CDA) in simulation

LLM-Driven Traffic Trajectory Generation with Game-Theoretic Modeling

• Game-theoretic trajectory generation for interactive multi-agent behaviors, validated on downstream perception and prediction tasks

Data-Efficient Perception under Sensing Uncertainty in Autonomous Driving

• Dataset optimization framework that reduced dataset size by 76.4% across four datasets while preserving 99.2% of model performance

• 1st Place, 2026 IEEE ITSS AV Challenge (Automated Vehicle Trajectory Modeling), ranked first among 28 teams

• The Best Graduate Thesis Award, Southern University of Science and Technology
• Highest Honors in Computer Science and Engineering, Southern University of Science and Technology
• Best Oral Presentation, 2025 IEEE International Conference on Computer Communication and Artificial Intelligence (CCAI)

• National First Prize, 2023 China College Computer Contest (CCCC)
• Excellent Award, 19th Challenge Cup of Science Achievement in China

• Reviewer, Transportation Research Board (TRB) Annual Meeting & Transportation Research Record
• Reviewer, International Journal of Information Technology & Decision Making