| CPC G08B 21/182 (2013.01) [B64C 39/024 (2013.01); G06F 18/24 (2023.01); G06T 7/97 (2017.01); B64U 2101/30 (2023.01); G06T 2207/10032 (2013.01); H04N 7/185 (2013.01)] | 6 Claims |
|
1. A method for detecting foreign object debris (FOD), the method comprising:
flying an unmanned aerial vehicle (UAV) on a specific flight path near a movement area surface at a specific speed and altitude, the UAV comprising avionics, an airframe, a location sensor, an attitude sensor, an edge processor capable of executing a machine learning model, at least one gimballed camera sensor, electromagnetic interference (EMI) sensor, a passive intermodulation (PIM) sensor, a high-intensity radiated field (HIRF) sensor, and a data connection to the avionics and a ground-based data center;
executing a computer vision application which positions the at least one gimballed camera sensor at specific angles and collects a plurality of images at a plurality of specific locations;
comparing, at the computer vision application, a plurality of baseline images with the plurality of collected images to detect a plurality of anomalies, wherein the plurality of anomalies comprises FOD;
tagging each of the plurality of anomalies with metadata including a GPS location;
transferring the plurality of tagged anomalies to a convolutional neural network for evaluation and classification;
transmitting over the data connection anomalies exceeding a predetermined threshold of duplicate detections in a same geographic location on overlapping images to a cloud service;
sending a notification comprising the plurality of evaluated anomalies exceeding the predetermined threshold to an inspection application on a mobile device; and
generating an alert on the inspection application that FOD has been detected.
|