Best AI Master’s Thesis Award 2021
Fredrik Svanström, a previous student at the Master’s programme in Embedded and Intelligent Systems, has received the best AI Master’s Thesis Award from the Swedish AI Society, SAIS. Congratulations!
The thesis is about detection of unauthorised drones at for example airports. Fredrik Svanström designed and built an automatic drone detection system that utilises machine learning and sensor fusion, which means that data from several different sources are combined. Besides the common video and audio sensors, the system also includes a thermal infrared camera and a receiver for aircraft transponder data. All collected data used to train and validate the system is published in an open database.
“The result of Fredrik’s work can help others to train their drone detection systems. It can improve safety at airports and other places where drones should not fly unauthorised”, says Cristofer Englund, Adjunct Professor at the School of Information Technology and one of Fredrik’s supervisors.
"Writing the thesis was indeed enriching, but for me, the Robotics course was the real prime of the Master's programme. We started with a pile of Lego bricks and assorted electronic components and ended up with a fully autonomous robot being able to perform a highly specified task with a strict time limit. That really tied together all the theoretical parts of the first year”, says Fredrik Svanström.
Fredrik Svanström graduated in the summer of 2020 from Halmstad University’s Master’s programme in Embedded and Intelligent Systems. The thesis title is "Drone Detection and Classification using Machine Learning and Sensor Fusion". Fredrik’s supervisors during the thesis work were Cristofer Englund, Fernando Alonso-Fernandez and Eren Erdal Aksoy.
The award was given by the Swedish AI Society, SAIS, on May 25, 2021, with the following motivation:
"The thesis is in the areas of machine learning and presents a system for multi-sensor-based drone detection and classification as well as a drone detection dataset. The thesis is well written, comprehensive and technically sound, with interesting results, not least in terms of the practical feasibility of multi-sensor-based drone detection. The thesis also offers an interesting outlook and constitutes a good starting point for future work."
The winning thesis will be presented at the SAIS workshop 2021 in June.