PhD defence Esma Mujkic

Title: Safe and Reliable Autonomy for Agricultural Vehicles


Principal supervisor: Professor Ole Ravn, DTU Electrical & Photonics Engineering
Co-supervisor: Associate Professor Nils Axel Andersen, DTU Electrical & Photonics Engineering
Co-supervisor: Senior Project Engineer Martin Peter Christiansen, AGCO A/S
Co-supervisor: Managing Director Morten Bilde Leth, AGCO A/S

Evaluation Board
Associate Professor Evangelos Boukas, DTU Electrical & Photonics Engineering
Professor Hans Werner Griepentrog, University of Hohenheim
Professor Arno Ruckelshausen, University of Applied Sciences Osnabrueck

Master of the Ceremony
Professor Lazaros Nalpantidis, DTU Electrical & Photonics Engineering

According to the latest UN World Population Prospects, the world's population is projected to reach 9.8 billion by 2050 and 11.2 billion by 2100. In order to meet the growing food demand and provide sustainable agricultural growth, it is crucial to optimize the use of land, labour, and other inputs through technological progress, social innovation, and new business models. In the last decade, precision farming and autonomous machinery have become essential components of the future vision of farming. The next crucial step on the roadmap is the development of autonomous farming vehicles that allow safe and reliable operation and reactive planning supported by advanced environment perception.

As it drives in the field, an agricultural vehicle must also continuously carry out a task in an environment that is highly unstructured and diverse. Therefore, developing a robust and accurate perception system is critical for autonomous agricultural vehicles to operate safely and reliably. The vehicle's ability to perceive and understand its environment relies heavily on processing visual inputs collected by cameras mounted on the vehicle. The recent success of deep learning architectures in computer vision tasks such as image classification, semantic segmentation, and object detection inspired the application of deep learning in solving scene perception tasks for autonomous vehicles.

The PhD study aimed to investigate further the real-life application of deep learning algorithms in scene perception in an agricultural environment. The project researched the application of state-of-the-art architectures for semantic segmentation, object detection and anomaly detection in scene understanding for agricultural vehicles. Additionally, the project proposed a framework for environment perception that combined the detection results of individual architectures to create a more comprehensive environment representation. 


tor 06 okt 22
13:00 - 16:00



The defence will take place in Building 303A, auditorium 49 at DTU Lyngby Campus and through zoom.