Technique of windshield de-icing state prediction based on neuronal model

Résumé

In this paper a technique of windshield de-icing state prediction is proposed. Parameters which affect de-icing performance in case of extreme cold conditions
have been taken into account to build a data base that is used to train a neuronal model. The obtained neuronal model will be used to predict de-ice state to warn the driver not to set the car into motion and allow the heating system to further warm the car interior and assure very good visibility on windshield. The proposed system will further enhance ADAS (Advanced Driver Assistance Systems) to improve traffic safety.