The project of an autonomous drone Starling is developed by a group of students. Some of them know each other already from a Warsaw secondary school, where they pursued their drone-related passions in the team RaByte. The present team met at the Warsaw University of Technology among the WUT MakerSpace community in October 2024. They were brought together by an ambition to compete with the best teams and by their interest in AI, computer vision and drone flights.

Innovation and technological challenge

At present they are taking part in the A2RLxDCL Autonomous Drone Race, co-organized by the Drone Champions League – a leading and most prestigious drone race league in the world. Within the competition, the teams must develop software that allows the drone to autonomously cover the route through defined gates in the shortest possible time.

The main task is to develop software for the drone so each team works on the same hardware. The drone has a monocular camera, which allows it to detect gates and determine location in space. It is supported by an IMU unit (Inertial Measurement Unit), which measures acceleration and angular velocities. A key role in data processing is played by an onboard computer Jetson Orin, which integrates information from various sensors, enabling precise drone control.

– The greatest challenge are solutions for automatic control, or actually training neural network for such control – says Jakub Maliszewski, Starling team leader. – For this purpose we used the motion capture system, worth over 200 thousand PLN, from the Faculty of Power and Aeronautical Engineering of the Warsaw University of Technology.

Leading academic teams from all over the world take part in the challenge, from for instance: UC Berkeley, TU Delft, Istituto Italiano di Tecnologia, University of Science and Technology of China and Korea Advanced Institute of Science and Technology. 

This is merely the beginning

After going through two qualifications stages, the Starling team will take part, already in April this year, in the great final of A2RLxDCL Autonomous Drone Race. For now they defeated over 50 teams and they believe their chances are high.

– We are already one of the 12 best teams among a few hundred teams – says Adam Chełchowski from Starling. – It is a source of great satisfaction and motivation for further work. And we have no shortage of ideas.

Team members are now planning further development, especially in research on drone groups, so-called swarms. This is a very future-oriented area. Autonomous drones are more and more often used not only in the military but also in areas like cartography, planning and logistics.

– Soon we are going to continue research which may be the basis for implementation of the new technologies – says Jakub Miętki. – We are considering enhancing our cooperation into a student research group, and in the long term, even starting a spin-off, which will allow us to implement our projects into production.

Why Starling?

This was inspired by the name of the bird. Starlings are very intelligent, which is proved by their acrobatics in the air. After the end of the reproduction period, they gather in groups of even a few thousand birds. When such a huge group of birds takes off, there is an amazing view in the sky. Sudden changes of direction, change of arrangement and other acrobatics make a huge impression. From a distance, the group resembles a black cloud, moving unnaturally fast and shifting its shape. This phenomenon is called murmuration. It is so perfectly synchronised that it is hard to believe that birds do not collide.

Starling team

The Starling team comprises: Jakub Maliszewski (team leader, student), Jakub Miętki (Data Science student), Jakub Twardowski (Electronics student), Adam Chełchowski (doctoral student at the Faculty of Power and Aeronautical Engineering), Mikołaj Markowski (graduate of industrial design at the School of Form), Antoni Rogowski (Automation, Robotics and Industrial Informatics student) and Michał J. Kozicki (Computer Science and Information Systems student).