facebook tracking

Master Thesis - Machine Learning Meets Localization

Explore the self-driving vehicle domain in this master thesis where you get to work with real life data!
Scroll to content


One of the first steps for any autonomous vehicle to drive by itself, consists of sensing its environment, localizing itself, and building a model of the road around it. One way to build an accurate road model is to use a High-Definition map and to accurately position the vehicle within the map. 

While HD-map localization can take advantage of the map content to provide better vehicle pose estimates, they should only be used to steer the car if we are certain to be at the right location. For example, localizing the vehicle in the wrong lane can have severe consequences. To avoid this, fast and robust initialization is important. 

One advantage of machine/deep learning is the ability to improve over time and over the amount of data. We want to explore different machine learning or deep learning approaches to improve the initialization of the vehicle pose, by utilizing both the onboard vehicle sensors (e.g. camera, radar, GNSS, lidar, etc) and map data from a large number of driving logs.

Project Description 

In this master thesis project, you will: 

  • preprocess raw data (from Zenseact's driving logs)
  • perform a literature review and select one or several machine-learning-based algorithms for vehicle pose initialization
  • train and evaluate the performance of the selected algorithm on the data 
  • compare accuracy and efficiency against our benchmark


We are looking for 2 students, preferably with good knowledge of

  • machine-learning/deep-learning
  • experience with some machine-learning frameworks (TensorFlow, PyTorch, scikit-learn, Keras, etc..) 
  • python programming 
  • knowledge regarding sensor fusion is considered as a plus.

Further information

Please send in individual applications with CV, motivational letter, and grade transcripts. 

Planned start: January 2022, with some flexibility.

Final application date: 15 of November 2021, but we will screen candidates continuously, so please submit your application as soon as possible.

Duration: 30 ECTS 

For questions regarding the project, please contact: 



Additional information

  • Remote status

    Flexible remote

Or, know someone who would be a perfect fit? Let them know!

Gothenburg, Sweden

Lindholmspiren 2
417 56 Gothenburg, Sweden Directions View page


Career site by Teamtailor