With the recent advances in robotics and artificial intelligence, human-robot interaction has become increasingly important for various applications. Autonomous robots are required to detect humans, provide safety, and follow them in some cases. Several approaches have been developed for detecting and tracking humans. In this paper, we propose a new approach that fuses two different human detection and tracking algorithms, namely the RWTH Upper Body Detector and Joint Leg Tracker, to provide a more robust human detection. We then use nearest-neighbor data association and Extended Kalman Filter to track the human. We also modify and integrate the Follow the Gap Method obstacle avoidance algorithm into the human tracking task to ensure a safe path and heading angle towards the person. Our proposed approach uses a similar approach to Adaptive Cruise Control for the robot to follow the person from a desired distance. We evaluate the performance of the proposed approach by considering the tracking distance error and heading angle difference between the human and robot in different scenarios.