A rough-fuzzy perception-based computing for a vision-based wall-following robot

This paper presents a new perception-based computing approach in a wall-following algorithm. The proposed perception-based computing uses a rough-fuzzy theory, which is an extension of the conventional fuzzy-based control approach. In practice, an indoor robot follows a wall in a compacted and complex environment with limited acquired data. Furthermore, visual sensor measurements may contain errors in a number of situations. In order to improve uncertainty reasoning results, it is necessary to perceive the encountered environment and filter the measured data. Therefore, a rough set theory is integrated to extract essential features of data to regulate inputs before applying fuzzy inference rules. The proposed control algorithm demonstrates excellent results through simulation and implementation.