An academic study examines two big data applications to manage multiple drones in swarms. The Drone Chasing Drones: Reinforcement learning and deep search area proposal published in July 2019 looks at two approaches which support cooperation and pursuit-evasion for Unmanned Aerial Vehicles (UAVs). The first uses vision-based deep learning object detection and reinforcement learning to detect and track a drone by another drone. It relies on a deep convolutional neural network to extract the target pose based on the previous pose and the current frame. The second approach uses a deep object detector and a search area proposal (SAP) to predict the position of the target UAV in the next frame for tracking purposes. This relies on historical detection data from a set of image sequences inputs this data to a SAP algorithm in order to locate the area with a high probability UAV presence. The aim is to develop architecture capable of tracking moving targets using predictions over time from a sequence of previously captured frames.
The study finds both approaches are promising and lead to a higher tracking accuracy overall. The study also finds that the deep SAP-based approach improves the detection of distant objects that cover small areas in the image. The researchers demonstrated their findings in outdoor tracking scenarios using real UAVs to test the proposed algorithms.
The study is published by MDPI, an open-access publisher for academic communities. It is compiled by Moulay A. Akhloufi, Sebastien Arola and Alexandre Bonnet, Perception, Robotics, and Intelligent Machines (PRIME), Department of Computer Science, Université de Moncton, Canada and UPSSITECH, Toulouse, France.
For more information visit:
www.dronebelow.com/2019/08/06/drones-chasing-drones-using-deep-learning-and-ai/
