RESEARCH DIRECTIONS

Current Research Contents

• Traffic Flow Oscillation and Shockwave Mitigation
  The team constructed a 1/10 scale RC car experimental platform to reproduce and mitigate the stop-and-go classic traffic phenomenon.

• Unsupervised Anomaly Detection in Driving Behavior
  The team proposed a Utility-based GRU-AE-GMM framework to address the challenges of current abnormal driving behavior detection, such as label scarcity and dynamic interactions.

• A High-resolution Vehicle–pedestrian Trajectory Dataset
  The team developed CrossTraj, a high-resolution vehicle–pedestrian trajectory dataset with a robust multi-camera processing pipeline to address the challenge of obtaining accurate, continuous, and real-world interaction data for traffic behavior modeling and safety analysis.

• Evaluation Frameworks for Roadside Perception Systems
  The team proposed a structured evaluation framework for roadside perception systems to address the limitations of generic metrics by enabling more accurate and robust assessment under real-world conditions such as distortion, occlusion, and synchronization errors.

• Two-Dimensional Surrogate Safety Measure
  The team proposed a two-dimensional surrogate safety measure (SSM) model that accounts for communication delay, packet loss, and detection errors to address the limitations of traditional SSMs in realistically evaluating the safety performance of connected and automated vehicles.

• Game Theory Decision-Making for Vehicle–Pedestrian Interaction
  The team proposed a discrete-time Markov decision framework based on game theory to enhance the safety of vehicles interacting with pedestrians at intersections.