Shape matching based leader tracking for a mobile robot with night vision ability

Abstract

One of the problems that a visually guided mobile robot has to deal with is night vision. Another application field of mobile robots is leader tracking. In this study, Bilecik Mobile Robot (BLM rho) which is designed for these purposes is introduced. While night vision is provided by an IRLED camera, shape carrying leader detection and following is achieved with regional descriptors, 2D cross-correlation and shape context. A more efficient following algorithm is obtained by decreasing computational cost with Kalman filtering. BLM rho takes the position of the center of the shape on the leader obtained from visual feedback and the shape area defining the distance between BLM rho and the leader as inputs. A fuzzy logic unit uses these inputs to generate control signals for left and right wheels. In the study, these stages are given in details, the methods used for leader tracking are compared and experimental results are presented.