Detect-and-Avoid (DAA) capabilities are critical for safe operations of unmanned aircraft systems (UAS). This paper introduces, AirTrack, a real-time vision-only detect and tracking framework that respects the size, weight, and power (SWaP) constraints of sUAS systems. Given the low Signal-to-Noise ratios (SNR) of far away aircraft, we propose using full resolution images in a deep learning framework that aligns successive images to remove ego-motion. The aligned images are then used downstream in cascaded primary and secondary classifiers to improve detection and tracking performance on multiple metrics. We show that AirTrack outperforms state-of-the art baselines on the Amazon Airborne Object Tracking (AOT) Dataset. Multiple real world flight tests with a Cessna 182 interacting with general aviation traffic and additional near-collision flight tests with a Bell helicopter flying towards a UAS in a controlled setting showcase that the proposed approach satisfies the newly introduced ASTM F3442/F3442M standard for DAA. Empirical evaluations show that our system has a probability of track of more than 95% up to a range of 700m. Video available at https://youtu.be/H3lL_Wjxjpw .
AirTrack: Onboard Deep Learning Framework for Long-Range Aircraft Detection and Tracking
2022
7 pages, 5 figures, ICRA 2023
Preprint
Electronic Resource
English
AIRTRACK, ein System fuer die Flugverkehrskontrolle
Tema Archive | 1977
|Airtrack: new rail access for London Heathrow Airport
British Library Conference Proceedings | 2008
|Deep learning of fault detection in onboard automobile systems
European Patent Office | 2024
|DEEP LEARNING OF FAULT DETECTION IN ONBOARD AUTOMOBILE SYSTEMS
European Patent Office | 2021
|