More than 1 million people die from traffic accidents each year around the world, which has arose public concerns to the vehicle safety. This paper examines improvement plan on the existent system of avoiding automobile rear-end collision. Based on the distance measurement millimeter-wave radar, the paper make predictions about relative distance and speed of the vehicle and the external substrate by principle of Kalman filtering. In order to predict the real distance of the vehicle and the outer, the author will correct the predicting process to the optimal estimation by comparing the predicted value with the observed value. That can improve vehicle safety and early warning mechanism accuracy.