A Co-Operative Perception System for Collision Avoidance Using C-V2X and Client-Server-Based Object Detection.

Area of Science:

• Intelligent Transportation Systems
• Wireless Communication
• Computer Vision

Background:

• Cellular Vehicle-to-Everything (C-V2X) communication, enabled by 5G, promises enhanced road safety through real-time data exchange.
• Existing C-V2X applications face challenges in real-time latency and communication reliability, particularly for non-equipped objects.

Purpose of the Study:

• To enhance safety at intersections and crosswalks by integrating C-V2X technology with environmental perception systems.
• To develop a co-operative perception system for effective collision avoidance.

Main Methods:

• Proposed a multi-module architecture combining C-V2X, Deep Neural Networks (DNNs) for object detection, and a client-server framework.
• Implemented GPS mapping for detected objects, ensuring accurate data for collision avoidance, even for non-C-V2X devices.
• Evaluated the system through real-time experiments on a dedicated testing track.

Main Results:

• The co-operative perception system achieved an average latency of 9.24 milliseconds for broadcasting critical obstacle information.
• Demonstrated accurate provision of GPS coordinates for detected obstacles, crucial for collision avoidance.
• The integrated system offers high data rates, low latency, and reliable communication.

Conclusions:

• The proposed C-V2X integrated system significantly enhances situational awareness and road safety at intersections and crosswalks.
• The system's ability to provide accurate GPS data for all detected obstacles is vital for effective collision avoidance.
• The technology integration is highly suitable for C-V2X applications requiring high performance.