Master Thesis project description

Predicting RSC in front of a moving vehicle using ML

The aim of this thesis is to predict the road surface condition (RSC) in front of a moving vehicle using machine learning (ML). Two kinds of data are available for this task: front-facing camera images and point-wise RSC estimates. The latter are measured underneath the vehicle, but they have been fused with the images through motion estimation, resulting in a sparsely annotated image segmentation dataset.

We see two potential focus areas for this thesis:

Annotation sparseness:
The RSC annotations are points that form lines going forward into the image where the vehicle has driven. However, the goal is to predict RSC across the entire roadway and not only along these lines. Can the sparse annotations somehow be augmented or extended to allow an ML model to learn to predict RSC outside of these sparse annotations?

Online learning:
ML models are typically trained once and then deployed in a frozen state. For this application, the RSC estimates could be used to update the model after deployment. How can the stream of RSC estimates arriving in real-time best be used to update the model?

Note:

We appreciate your interest in Klimator. Klimator is an equal employment opportunity employer, committed to maintaining a supportive and inclusive workplace for all employees. Klimator does not discriminate against, and prohibits harassment of, any applicant or employee based on race, color, sex, sexual orientation, gender identity, religion, national origin, age or disability.