Autonomous Non-Destructive Inspection

We offer a framework that will provide flexible automatic programming of robots and a vision system used for matching unknown structural components to known structural components with a machine learning engine and a training library of the types of components which will be encountered. Our library includes a parameterized scan path for each component and the inspection technique we used (Eddy Current, Through-Surface Transmission, Thermography, etc.) When an unknown component is encountered, we extract the relevant features from its CAD model (bounding box size, surface area, orientation, curvature, etc.)  and use machine learning to match it against our library of known components to find the probability of a match. Our library of parts is selected as the representative components of those we will encounter, and the scan paths are taught by an experienced NDI programmer using our industry hardened E-HUB NDI software. This information is stored in a format that will make machine learning matching viable.