Tyre-road friction correlation to enable autonomous driving at all weather and road conditions

Important results from the project

The autonomous vehicles of the future will depend on precise information about the available friction between tyres and the road surface. Today´s friction estimators are not good enough and machine learning has been seen as a possible way forward. This project has developed a method where road friction equipment is used to generate high-quality training data for data-driven development of friction estimators. Mappings for three different winter tyres have been developed for ice and snow roads conditions.

Expected long term effects

The results of extensive field tests on ice and snow show that a dedicated road friction meter can be a useful tool for continuously measuring the available braking friction on winter roads for a specific vehicle and tires. The three mappings developed for different types of winter tires will be useful for generating data for machine learning in order to improve existing friction estimators for passenger cars, and by extension autonomous vehicles.

Approach and implementation

The method development has been based on a commercial road friction meter and via test track measurements determine the correlation between longitudinal friction values from this device and a target vehicle, equipped with different winter tires at different winter road conditions. From extensive tests on ice and snow tracks with these equipment, together with VTI´s tire testing equipment, a cost-effective open method for performing mappings of this kind has been developed.