Friction Potential Estimation at low levels of Friction Consumption using 1D Convolutional Neural Networks

This thesis focussed on friction potential estimation at low levels of excitation with deep neural networks.

Completed Master Thesis

For every wheeled-propelled system, being able to estimate the friction potential is an element that can improve performance and safety. However, the task is not trivial, especially at lower levels of excitation.

The goal is to find a Deep Neural Network that analysing time series data, provided by standard on board sensors, can give a reliable friction potential estimation.

The Network will be developed in collaboration with Daimler. The data used for the  training and testing process are collected by a real-life vehicle.


To the top of the page