A driver adaptive lane departure warning system based on image processing and a fuzzy evolutionary technique

By applying fixed threshold values for a single lane departure information such as lateral offset, most lane departure warning systems have a difficulty in practical application. This paper proposes a method that can generate the driver adaptive lane departure warning model that can reduce driver disturbance factors, such as a frequently annoying alarm, etc. After a training period, the lane departure warning model is constructed by a fuzzy-evolutionary algorithm that can fuse the current and near future vehicle state information like the lateral offset and TLC (Time to Lane Crossing). After the departure model has been constructed; the driver merely selects an appropriate hazard level of lane departure warning. The proposed system has been developed and tested in HiLS (hardware in the loop simulation).