Case study: robotic person following using fuzzy control and image segmentation

Abstract

Abstract The paper proposes a scheme for amobile robot to follow a person in reasonably complex dynamic environments. The scheme consists of an image processing technique for person identification, and a fuzzy controller for following the person. The identification process utilizes image color, shape characteristics, and a region growing technique. The fuzzy logic controller receives image data such as the mass, center of mass, and their rates of change to produce steering and speed commands for the robot. A behavior control strategy is also incorporated to cope with several situations, such as hazard, that are not accounted for in the fuzzy controller. Descriptions of the software/hardware as well as experimental setup and results are provided to demonstrate the feasibility of the scheme. © 2003 Wiley Periodicals, Inc.