Swiss Engineering Ticino Prize 2020 to Marco Calzana

The award ceremony was held on 24 September at Centro manifestazioni mercato coperto Mendrisio.
The award ceremony was held on 24 September at Centro manifestazioni mercato coperto Mendrisio.
The award ceremony: from the left the representative of the Faculty of Informatics Mauro Prevostini, the winner Marco Calzana, the president of the jury Enrico Vitali and the President of SwissEngineering Ticino Monica Gianelli Bertino.
The award ceremony: from the left the representative of the Faculty of Informatics Mauro Prevostini, the winner Marco Calzana, the president of the jury Enrico Vitali and the President of SwissEngineering Ticino Monica Gianelli Bertino.

Institutional Communication Service

25 September 2020

The PremioSwissEngineering Foundation aims to promote the work of Ticino's new graduates in the field of engineering and to financially support young talents from the vocational schools and universities of Canton Ticino. The prize for the category "USI Faculty of Informatics" has been awarded to Marco Calzana for his Master's thesis on a topical issue with very concrete potential: the improvement of reliability and the experience of independent driving.

Marco Calzana earned the Master in Software & Data Engineering (USI Faculty of Informatics), at the Software Institute. His thesis "Misbehaviour prediction in autonomous vehicles using autoencoders" has been written under the supervision of Prof. Dr Paolo Tonella and Dr Andrea Stocco. 

Marco chose to focus his thesis on an important goal with concrete implications: to detect in advance anomalous situations that may occur in autonomous vehicles using autoencoders. In the case of probable collisions or road exits, the system enables a self-correction. In essence, the thesis answered this research question: "Can a self-driven car predict if it is faced with an unexpected situation that will lead to a violation of the requirements? To answer this question, in his research, Marco used and perfected a simulator that uses Deep Neuronal Network techniques to predict in advance unexpected autonomous driving behaviour.

«Worth emphasising is the degree of innovation of this thesis»- says the laudatio pronounced on the occasion of the award ceremony - «which is highlighted by the fact that the system can predict incorrect behaviour of an autonomously driven car with a degree of reliability of around 77%, rather than detect it after it has occurred. The results obtained in the field of improving the reliability and the experience of autonomous driving, let us imagine the interesting potential in a cutting-edge sector with a promising future». 

 

Faculties

Sections