Road surface condition detection with recursive adaptive learning and validation

A method for determining a road surface condition for a vehicle driving on a road, the method comprising the steps of:

• determining, by a first probability module, probabilities associated with a plurality of road surface conditions based on an image of a capture scene of the road;
• determining, by a second probability module, probabilities associated with the plurality of road surface conditions based on vehicle operating data;
• inputting the probabilities determined by the first probability module and probabilities determined by the second probability module to a data fusion and decision making unit for fusing the probabilities determined by the first and second probability modules;
• determining a road surface condition based on fused results by the data fusion and decision making unit;
• outputting a refined probability associated with the identified road surface condition from the data fusion and decision making unit that is a function of the fused first and second probabilities;
• and
• providing the refined probability from the data fusion and decision making unit to an adaptive learning unit, the adaptive learning unit generating output commands that refine tunable parameters of at least the first probability module and the second probability module for determining the respective probabilities.