Below Horizon Aircraft Detection Using Deep Learning for Vision-Based\n Sense and Avoid

2019/03/07 by Jasmin James, James, Jasmin, Jason Ford +3
Computer Science · Engineering · #Advanced Neural Network Applications #Air Traffic Management and Optimization #FOS: Computer and information sciences #Robotics (cs.RO) #UAV Applications and Optimization

paper · pdf · doi:10.48550/arxiv.1903.03275

openalex publication_date 2019/03/07 · openalex created_date 2022/07/29 · openalex updated_date 2026/07/28

Abstract

Commercial operation of unmanned aerial vehicles (UAVs) would benefit from an\nonboard ability to sense and avoid (SAA) potential mid-air collision threats.\nIn this paper we present a new approach for detection of aircraft below the\nhorizon. We address some of the challenges faced by existing vision-based SAA\nmethods such as detecting stationary aircraft (that have no relative motion to\nthe background), rejecting moving ground vehicles, and simultaneous detection\nof multiple aircraft. We propose a multi-stage, vision-based aircraft detection\nsystem which utilises deep learning to produce candidate aircraft that we track\nover time. We evaluate the performance of our proposed system on real flight\ndata where we demonstrate detection ranges comparable to the state of the art\nwith the additional capability of detecting stationary aircraft, rejecting\nmoving ground vehicles, and tracking multiple aircraft.\n

Related

• A Novel Technique for Rejecting Non-Aircraft Artefacts in Above Horizon Vision-Based Aircraft Detection
• AirTrack: Onboard Deep Learning Framework for Long-Range Aircraft Detection and Tracking
• A Dataset of Stationary, Fixed-wing Aircraft on a Collision Course for Vision-Based Sense and Avoid
• Ensuring UAV Safety: A Vision-only and Real-time Framework for Collision Avoidance Through Object Detection, Tracking, and Distance Estimation
• Drone Detection Using Depth Maps