A study on warning activation timing of rear-end collision with driver model of neural network

This study constructs a driver model that reflects the braking characteristics of individual drivers. In addition, the purpose of this study is to establish the effectiveness of applying the said driver model to the activation timing of the collision warning system. The experiment was carried out using a driving simulator, and evaluation of the system's activation timing was carried out from both subjective and objective perspectives. For system rating on the effect of collision warning system using the driver model, it was compared with a system with the activation timing set uniformly based on the technical guidelines. As a result, following conclusions were reached. (1) If the deceleration of the forward vehicle is significant, driver's acceptance is not lost even if the collision warning system using the driver model is activated early. (2) The collision warning system using the driver model increases the durational margin until collision, and contributes to accident avoidance with driver action. (3) The collision warning system using the driver model can reduce the risk of collision with forward vehicles and improve the safety level. In addition, the load of driver’s braking action can be reduced.