Localization and obstacle detection for robots for carrying food trays

This paper discusses a practical system which, when incorporated in an autonomous mobile robot moving indoors, can determine the position of the robot and detect obstacles so that the robot can navigate appropriately. We have developed new real-time, reliable techniques using correlation operations to achieve vision-based localization and obstacle detection. These techniques include a landmark detection technique that assures stable detection even under variable brightness and an obstacle measurement technique that combines obstacle region segmentation with stereo vision. Our experiments have proven that these techniques can make the necessary measurements in 100 ms, assure stable measurement in environments of varying brightness, and are sufficient for mobile robots.