Advance visual inspection system (ADVISE): A prototype inspection workstation that supports remote visual inspection of aircraft surfaces

Direct visual inspection is the most widely used method employed for finding surface defects on aircraft surfaces. It requires placing a human inspector on the body of the aircraft to visually examine its surface for defects. An alternative approach to direct visual inspection is robot-based remote visual inspection. Here, an aircraft-capable mobile robot is placed on the body of the aircraft and is controlled remotely by an inspector on the ground. The inspector performs inspection by examining live video of the aircraft surface captured and relayed to the ground by the robot.;This thesis describes a prototype computer-based inspection system called the "Advance Visual Inspection System" (ADVISE) developed to facilitate robot-aided remote visual inspection. ADVISE acts as an interface between the robot on the aircraft body and the inspector on the ground. It is a platform for displaying live video relayed by the robot as well as providing tools for computer-aided video enhancement and automated defect detection. Some of the key subjects that the thesis addresses are: (1) Crown Inspection Mobile Platform (CIMP), a mobile-robot developed at the sensor lab at Carnegie Mellon Robotics Institute; (2) Library of monoscopic and stereoscopic video enhancement algorithms; (3) Automated surface crack and surface corrosion detection algorithms; (4) The role of ADVISE in remote visual inspection.;The thesis provide insight into the design challenges and design strategy adopted in the development of the enhancement and automated surface defect detection algorithms. Various representation models used by modules of these algorithms are described in detail. The algorithms are evaluated through a quantitative analysis of its performance on real aircraft surface defects. Finally, a discussion on how these algorithms can be used in robot-based remote visual inspection is included in the thesis.