Many service robots and various service modules are being developed. Robots for services, such as guiding, escorting, and nursing, require human-friendly navigation, which is an appropriate objective of interaction. This paper proposes a method for a mobile robot to detect a human leg and to follow the human for interaction with him/her. Most human-tracking schemes in mobile robots utilize vision sensors and laser range scanners. Generally, mobile robots confirm the distance to a human through a laser range scanner after recognizing the human by vision. Our approach clearly reveals the attributes of a human leg through one laser range scanner which is used to detect obstacles and localize. Using attributes that have been experimentally obtained by the accuracy range data with a laser range scanner; we have striven for robots to be able to distinguish legs in their environment. We have applied our suggested walking model for the case where mobile robots lost track of the target person's legs and failed to detect those legs.