Improved accuracy and robustness in drone localisation: Award of Excellence for Alessandro Fornasier

Alessandro Fornasier’s work involves helping robots and drones to navigate the world. As part of his PhD project, he used the theory of ‘equivariant systems’ to develop localisation algorithms that are more robust and accurate. In recognition of his research in the Control of Networked Systems working group, he has now been honoured with the ‘Award of Excellence’ presented by the Austrian Federal Ministry of Education, Science and Research.

“Our aim was to apply the mathematics of symmetry to improve the current state-of-the-art in determining the orientation, position and speed of drones”, Alessandro Fornasier explains.

The drones are usually equipped with various sensors and collect data. So, for example, if the drone moves from A to B within one second, and collects camera and GPS data, it can use this to derive location information. However, here lies the problem: Wrong sensor information, as well as sensor biases, yields inaccuracies and a tendency to make mistakes when determining the location, especially with increasingly agile drones. “We want to safeguard the drone, and develop methodologies that are robust against such errors. Even if individual information are wrong or biased, it should still be possible to calculate a the correct result for the alignment, position and speed,” Alessandro Fornasier continues.

Alessandro Fornasier was able to successfully close this gap by relying on the ‘Equivariant Filter’ method, which is based on a theory developed by Robert Mahony (Australian National University). Many physical systems carries natural symmetry. Exploiting those symmetries is of paramount importance to design algorithms that can describe – and subsequently minimise – the orientation, position and speed error. When Alessandro Fornasier and his supervisor Stephan Weiss (Control of Networked Systems research group) decided to explore the localisation problem at this fundamental level, there was no guarantee that it would have worked. In the three years that followed, Fornasier worked intensively on this issue, joined, among others, by his eventual co-supervisor, Robert Mahony. It worked: “We were able to show that the state of the drone can be estimated more accurately and robustly using this methodology,” says Alessandro Fornasier. “This means we have created a promising basis that can be adopted and refined by others. The gap between the theory of symmetry and the problem of state estimation has now been closed.”

Alessandro Fornasier’s fascination with drone research began in 2019, when he transferred to the University of Klagenfurt from the Università degli Studi di Udine as part of his double-degree Master’s programme in ‘Information and Communications Engineering, Autonomous Systems and Robotics’ and met Stephan Weiss and Jan Steinbrener in the Control of Networked Systems research group. “They helped me to become a mature researcher. They always pushed me to dig deeper into the nature of research problems and get an understanding at the fundamental level”, he says, looking back. Alessandro Fornasier completed his doctoral studies in June 2024. In December 2024, he was honoured for his outstanding work with the ‘Award of Excellence’ presented by the Federal Ministry of Education, Science and Research. Alessandro Fornasier has since settled in Zurich, where he works on innovative technologies for robot localisation at Hexagon Robotics Hub.