Computer Vision for UAV Swarm Traffic Monitoring

Supervisor: Univ.-Prof. Dr. Radu Prodan
Author: Dominik Schweigl

Abstract

This bachelor thesis aims to develop and evaluate a computer vision algorithm for traffic monitoring using a UAV (Unmanned Aerial Vehicle) drone swarm. The proposed system serves as a proof of concept for integrating UAV swarms into intelligent transportation infrastructure, with the goal of identifying and analyzing dynamic traffic phenomena such as congestion or moving bottlenecks in hard-to-access areas. The focus lies in developing a detection algorithm that utilizes optical, thermal, and distance measurements for traffic pattern recognition. The work may also include an overview of sensor fusion techniques and an exploration of edge computing strategies to execute the analysis algorithm across multiple UAVs. Optionally, the project can be extended by comparing CNN- and Vision Transformer-based architectures or by evaluating the effectiveness of different sensor combinations (optical, thermal, distance) under multiple environmental conditions such as day and night.