The traditional pure pursuit controller is often difficult to deal with the control delay brought by the sensor signal delay, actuator response lag and other factors, as well as the lack of active anti-disturbance ability under the intervention of internal and external disturbances. At the same time, the preview distance is used as the denominator of the control law and it gradually will decrease when tracking the target point near the end, resulting in sudden changes in the input angle value and along with the unstable steering wheel, which affects the driving stability and the riding experience of the passengers. A path tracking controller based on improved pure pursuit was proposed. Firstly, the base model of the pure pursuit controller was constructed and a road preview model combining the planned path and the planned speed information was established to improve the lagging effect of the pure pursuit control. Then, the unknown disturbances generated by the internal and external environment of the system were considered, a nonlinear expanded state observer (NESO) was designed to estimate and compensate the disturbances in real time to enhance the anti-disturbance capability of the pure pursuit controller. Furthermore, a steering stability adjustment model was developed to improve the sudden change of steering angle when it's approaching the end point. Finally, the effectiveness of the proposed controller was tested on a software-in-the-loop dynamics simulation platform with different initial heading and initial deviation conditions. Further, the proposed controller was verified in a closed-loop test on a real-world vehicle platform, and the experimental results show that improved pure pursuit algorithm has good tracking accuracy and steering smoothness.