This paper presents a GPS-based method for outdoor robots to track humans. This new method can overcome the crucial problems of conventional techniques in complex environments with obstacles or sloped terrain that do not allow detecting the locations of humans out of the robot"s line of sight. The robot determines the position of the human with respect to GPS data and forms the trajectory of the human’s movement. This trajectory is then smoothed in real time to reduce sudden changes in the path and improve the tracking performance. We also propose an autonomous trajectory tracking method for the robot to avoid obstacles while effectively tracking the human trajectories. This method allows the robot to follow the human even in an environment with many robots and humans simultaneously present because the robot can always distinguish the human it should follow. The experiments demonstrate that robots can effectively follow the human while avoiding obstacles in complex environments.