Digital Twin for Pedestrian Safety Warning at a Single Urban Traffic Intersection

Yongjie Fu2, Mehmet K. Turkcan3, Vikram Anantha4, Zoran Kostic3, Gil Zussman3, Xuan Di2,1
2Department of Civil Engineering and Engineering Mechanics, Columbia University, New York, NY, 10027.
3Department of Electrical Engineering, Columbia University, New York, NY, 10027.
1Data Science Institute, Columbia University, New York, NY, 10027.
4Lexington High School, Lexington, MA, 02421.
Paper Code

Real-time demonstration of our Digital Twin system monitoring pedestrian safety at an urban traffic intersection, providing instant warnings for potential safety hazards.

Abstract

Pedestrian safety at urban traffic intersections remains a critical concern in modern transportation systems. This paper presents a comprehensive Digital Twin framework for real-time pedestrian safety monitoring and warning generation at a single urban traffic intersection. Our system integrates computer vision techniques, sensor fusion, and predictive modeling to create a virtual representation of the physical intersection environment. The Digital Twin continuously monitors pedestrian movements, vehicle trajectories, and environmental conditions to identify potential safety hazards and generate timely warnings. We employ advanced object detection and tracking algorithms to maintain accurate real-time representations of all intersection participants. The system's predictive capabilities enable early warning generation before dangerous situations occur, significantly improving pedestrian safety outcomes. Experimental results demonstrate the effectiveness of our approach in reducing pedestrian-vehicle conflicts and enhancing overall intersection safety. Our framework provides a scalable solution that can be adapted to various urban traffic scenarios, contributing to the development of safer and more intelligent transportation systems.

Raw image output from high-altitude camera
Object detection results
Sample object detection results from the high-altitude camera deployment enabling the Application.
Left - Raw image output retrieved from the camera. Right - Object detection results.
System Architecture Flowchart
Flowchart for how the entire system architecture works.
CARLA Simulator Flowchart
Flowchart for how virtual pedestrian in CARLA simulator is involved.
3D model of COSMOS testbed in CARLA - street view
3D model of COSMOS testbed in CARLA - top view
3D model of COSMOS testbed in CARLA
Validation with virtual pedestrians in CARLA
Use virtual generated pedestrians to validate the safety warning algorithm.
TTC in CARLA simulation
Pedestrian coordinates in CARLA simulation
TTC (left) and pedestrian coordinates (right) in CARLA simulation

Paper

BibTeX

@article{fu2024digital,
  title={Digital Twin for Pedestrian Safety Warning at a Single Urban Traffic Intersection},
  author={Fu, Yongjie and Turkcan, Mehmet K. and Anantha, Vikram and Kostic, Zoran and Zussman, Gil and Di, Xuan},
  journal={[Journal/Conference Name]},
  year={2024},
  publisher={[Publisher]}
}