Abstract
Fully autonomous drones have been demonstrated to find lost or injured persons under strongly occluding forest canopy. Airborne optical sectioning (AOS), a novel synthetic aperture imaging technique, together with deep-learning-based classification enables high detection rates under realistic search-and-rescue conditions. We demonstrate that false detections can be significantly suppressed and true detections boosted by combining classifications from multiple AOS—rather than single—integral images. This improves classification rates especially in the presence of occlusion. To make this possible, we modified the AOS imaging process to support large overlaps between subsequent integrals, enabling real-time and on-board scanning and processing of groundspeeds up to 10 m/s.
Highlights
Autonomous drones have been demonstrated to find lost or injured persons under strongly occluding forest canopy
We have demonstrated the real-time application of Airborne optical sectioning (AOS) in fully autonomous and classification-driven adaptive search and rescue (SAR) o perations[31]
We demonstrate in “Combined classification” and “Results” sections that the product of median and maximum confidence scores of combined classifications significantly suppresses false while boosting true detections
Summary
Autonomous drones have been demonstrated to find lost or injured persons under strongly occluding forest canopy. Airborne optical sectioning (AOS)[23–31] is an effective wide-synthetic-aperture aerial imaging technique that can be deployed using camera drones It allows virtual mimicking of a wide aperture optic of the shape and size of the scan area (possibly hundreds to thousands of square meters) that generates images of an extremely shallow depth of field above an occluding structure, such as a forest. These images are computed by integrating regular single pictures that are captured by the drone and allow optical slicing through dense occlusion (caused by leaves, branches, and bushes). Approaches for measurement-based combination can be further categorized into a daptive[47–50] and non-adaptive
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
