Development of a Technology to Enhance Automated Driving Using Infrastructure Guidance

Introduction : Integrated Traffic Support System via Infrastructure–Vehicle Cooperative Driving in Urban Environments 

This research project aims to develop an infrastructure-cooperative autonomous driving service that enhances traffic efficiency and safety in complex road environments where Level 4/4+ autonomous vehicles, conventional vehicles, and vulnerable road users (VRUs) coexist. The core objective is to achieve comprehensive situational awareness and optimal driving guidance through cooperative perception between vehicles and infrastructure, ultimately realizing a safe and efficient integrated traffic system for all road participants. 

Dataset Strategy for Realistic and Diverse Traffic Scenarios 

To reflect diverse traffic environments, this study utilizes inD and rounD open datasets collected via drones to learn complex interactions at intersections. Scenario-based simulation data was generated using the K-City simulator to match domestic road conditions. Additionally, an infrastructure-focused dataset based on Edge RSU observations is being constructed to support real-time cooperative perception development. 

Deep Learning Approaches to Cooperative Driving Behavior Prediction 

To enable the proposed service, the research focuses on developing trajectory prediction and maneuver prediction algorithms. These models are designed using Transformer-based multitask architectures, allowing robust performance in complex traffic situations. By identifying failure cases, the team refines model performance through improved data preprocessing, loss function design, and training strategies, aiming for a more reliable and adaptive cooperative driving system.

System Testing and Deployment 

The developed technologies are being tested in K-City, autonomous vehicle testbed of Korea, to validate their real-world applicability. Additionally, the project is preparing for deployment in actual urban roads, including areas such as Namyang-eup and Saesol-dong in Hwaseong City, with plans to expand demonstrations to broader regions in the near future.