Abstract: To solve the problem that the traditional lane keeping system has inadequate control precision and the driver model has insufficient adaptability to the road with large curvature,an improved algorithm for the optimal preview acceleration driver model is investigated by Taylor expansion at non-zero frequency point.The driving simulator data and the fast Fourier transformation (FFT) method were used to study the main steering frequency under double lane change.And a Simulink/CarSim co-simulation model was built as well.Double lane change simulation experiments were carried out.The results show that the lateral displacement error of the improved model is significantly smaller than that of original one,which the maximum displacement errors are reduced by 0.2,0.15 and 0.11 m for vehicle speeds of 60,70 and 80 km/h respectively,and that the improved driver model can control the vehicle to follow the expected path better than the road artificial potential field method.

Key words: driver model, active safety, lane keeping, steering frequency, lateral control