Abstract：To make the intelligent vehicle obtain humanlike behavior when passing a curve road, a humanlike steering control driver model was proposed based on optimal curvature preview driver model and TS fuzzy inference system to adjust the preview point on both lateral and longitude directions. The driving data from experienced driver model were collected in an urban curve road, and the optimal target was obtained after data analysis. Ant colony optimization (ACO) was used in tuning the fuzzy parameters in the preview decision module. A multicurvature driving scenario was built in Prescan, and the humanlike characteristic of the proposed driver model was verified by the trajectories of experienced drivers collected by a driving simulator. The results show that the trajectory similarity between the experienced drivers and the humanlike steering driver model is higher than those of the optimal curvature preview model and the driver model with adaptive preview distance.
