Master's thesis: Knowledge Transfer and Federated Learning for Heterogeneous Object Detection Models in Autonomous Vehicles

Master's thesis: Knowledge Transfer and Federated Learning for Heterogeneous Object Detection Models in Autonomous Vehicles Apply here Background Autonomous vehicles rely on AI models trained on large-scale, multi-modal sensor data. As vehicle platforms evolve, changes in sensors and hardware often require updating or retraining these models, which is costly and time-consuming. A key challenge is therefore how to efficiently transfer knowledge between models operating under different configurations. This thesis is part of the research project DREAM – Distributed, Robust and Efficient AI for Autonomous Vehicles. The topic is highly relevant for enabling scalable and efficient AI development in next-generation autonomous driving systems. Description Sensor data and AI enable cars to detect objects, understand their environment and make decisions about how to respond. When vehicles are updated and new models are developed, sensors and hardware often change, which in turn also affects the AI models used. One approach would be to create a new AI model from scratch and collect new data each time the vehicle platform is updated. A more efficient solution would be to transfer knowledge between models with varying architectures. In this Master’s thesis project, we aim to investigate knowledge transfer between diverse models and hardware setups, ensuring that learning can continue even when architectures change. The work will use the Zenseact Open Dataset and also explore knowledge transfer in a federated learning context. Main Tasks In this master thesis project, you will focus on investigating knowledge transfer between diverse models and hardware setups in autonomous vehicles. Specifically, you will: Explore how knowledge can be efficiently transferred between AI models with different architectures. Evaluate techniques for updating AI models when vehicle sensors or hardware change, without the need for full retraining or collecting extensive new datasets. Analyze the effectiveness of the proposed approach through experiments using multi-modal sensor data, including vehicle control signals, geographical positions, and lidar, radar, and camera measurements. Develop and experiment with a federated learning framework that incorporates knowledge transfer to maintain model performance and adaptability in real-time, large-scale deployments. Qualifications We are looking for one or two highly motivated students with a good general background in machine learning and computer vision. The following skills would be essential: Deep learning Federated learning (would be a bonus) Python programming Reading scientific papers Handling complex systems Conditions Location: RISE, Kista, Stockholm Applications are reviewed on a rolling basis, apply as soon as possible, but no later than August 31st, 2026. Starting date: As soon as possible, not later than September 1st, 2026. Credits: 30 points Compensation: 39990 SEK upon a successful completion of a high-quality thesis. Supervisors: Henrik Abrahamsson (RISE) Sima Sinaei (RISE) Welcome with your application! Send in your application (CV, motivation letter, transcript of records) no later than August 31st. For any questions, please contact: Henrik Abrahamsson, [email protected] Sima Sinaei, [email protected] Category Student - Thesis Locations Stockholm Remote status Hybrid About RISE Research Institutes of Sweden AB RISE is Sweden’s research institute and innovation partner. Through our international collaboration programmes with industry, academia and the public sector, we ensure the competitiveness of the Swedish business community on an international level and contribute to a sustainable society. Our almost 3300 employees engage in and support all types of innovation processes. RISE is an independent, State-owned research institute, which offers unique expertise and over 130 testbeds and demonstration environments for future-proof technologies, products and services. Founded in 2016 Co-workers 3000+