sUAS Payload AI Recognition of Objects
Tuesday, October 06, 2020: 12:05 PM - 1:00 PM
We present a use case of an object detecting sUAS payload employing the use of a single deep neural network. The payload takes in a live video from sUAS’ perspective and performs detections at the edge. Throughout development, the objective was to create a robust system architecture that is not hardware or data input dependent. This approach led to the creation of two separate models that were integrated into a similar fashion to create the sUAS mountable payload. The specific use case was to train a Single Shot Detector to detect people and relay their geolocation data to a ground display. The DJI Matrice 600, a high performance and endurance sUAS, was selected to carry the payload for the DHS & SOCOM demos. Further ground testing was done in-house, and the object detection capability was presented at the AUVSI Hill Day demo in September.
Analyst,Engineering/Technical,Research & Development,Student