In this paper, a novel quadratic finite horizon cost function was proposed to solve the trajectory tracking problem of a wheeled mobile robot. This cost function accounts for the correlation of past heading angle state of the kinematic model of a wheeled mobile robot to smooth out the trajectory following behavior. A first-order polynomial is fitted to the past heading angles and future heading angles in the control horizon are predicted with this first-order polynomial; when the cost function is minimized with model predictive control the mobile robot makes a compromise between predicted heading angles and the future states on the reference path. It has been shown that with the use of the proposed model predictive control, the autocorrelation of the heading angle signal of a wheeled mobile robot is maximized. Thus, a smooth trajectory tracking behavior is achieved on the reference path.