Human-robot interaction is an important issue for service robots. In this paper, we propose to utilize the multisensory data fusion to autonomously detect and track an accompanist as a target person in the surroundings of the mobile robot. First, we use laser range finder to detect the target person features which is characterized using distances and orientations from range sensor readings. Based on the sensor information and the desired tracking position, a fuzzy based tracking controller is proposed to follow the accompanist. In addition, the obstacle avoidance strategy is also embedded into the following controller to mitigate disturbances or interferences in dynamical environment when robot is performing a following task simultaneously. If the accompanist is missed, a recognition process based on a pan-tilt-zoom camera and observed images will be used to search the target person again. Experimental results demonstrated the validation and feasibility of the proposed system.