Traffic Monitoring System for Unmanned Aerial Vehicles

My undergraduate research focused on the development of an embedded traffic monitoring system capable of detecting, tracking, and estimating the position and speed of vehicles from aerial imagery acquired by unmanned aerial vehicles (UAVs).

The project combined computer vision, machine learning, and embedded systems to perform real-time onboard image processing using a single-board computer integrated with a camera, GNSS receiver, and inertial sensors. Vehicle detection was implemented using Histogram of Oriented Gradients (HOG) descriptors and a Support Vector Machine (SVM) classifier, while optical flow techniques were employed for object tracking and motion estimation. Sensor fusion was then used to estimate the geographic position and speed of each detected vehicle.

Besides the software implementation, the project also included the design and integration of the embedded hardware platform, including sensor interfaces, power system, synchronization mechanisms, and installation on a fixed-wing UAV for experimental validation.

This work represented my first experience integrating embedded hardware, computer vision, sensor fusion, and autonomous aerial systems, establishing the foundation for my subsequent research in embedded systems, satellite engineering, and space technologies.

Keywords

Embedded Systems · Computer Vision · Machine Learning · UAVs · Image Processing · Optical Flow · Sensor Fusion · GNSS

Degree

B.Sc. in Electrical Engineering Federal University of Santa Catarina (UFSC), 2016

Advisor

Prof. Joceli Mayer