Facilitating Real-time Road Event Detection on Collaborative UAV and Ground Vehicle

Abstract

Smart road systems have undergone significant advancements, integrating technologies like the Internet of Things, Artificial intelligence, and big data analytics to enhance the efficiency, safety, and sustainability of transportation networks. Despite these developments, road accidents remain a pressing issue, underscoring the need for advanced road event detection systems. Traditional vehicle-mounted sensors (e.g., Lidar, Radar, and stationary cameras) each have limitations, such as restricted sensing ranges and high costs. Drones, with their flexible mobility and extended sensing capabilities, offer a promising solution for road event detection. However, their limited battery capacity and computational resources necessitate distributing the computing to external resources.This thesis proposes a novel approach for real-time road event detection through a collaborative system combining drones and ground vehicles (GVs). By leveraging the computational capabilities of GVs and the aerial advantages of drones, this framework aims to deliver fast, accurate, and energy-efficient event detection. Key contributions include the design and implementation of a distributed middleware system, Griffin, facilitating drone-car collaboration for real-time road event detection, and empirical evaluations demonstrating its effectiveness. The work concludes with a discussion of challenges and future directions for further enhancing drone-based road event detection systems.