RGB-D IMAGE BASED LANDMARK IDENTIFICATION FOR CORRIDOR COVERAGE NAVIGATION

Developing robotic coverage navigation system in straight corridor using only one sensor, RGB-D camera, is a challanging problem. System must be able to identify the landmarks that naturally exist in structured environment. This ability is needed to provide the information about perimeters of workspace that should be covered by robot in a single task at segmented area of corridor. Color detection and distance estimation using RGB-D images are applied to identify landmarks that found in corridor. Identification is calculated from salient region that covered 0.125 of images area. Landmarks in range between 0.8 to 5.6 meters have successfully identified with accuracy 98.86%.