Please use this identifier to cite or link to this item:
Title: Autonomous deep learning for unmanned aerial vehicle
Authors: Looi, Deane Yi Ren
Keywords: Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence
Issue Date: 2020
Publisher: Nanyang Technological University
Project: SCSE19-0081
Abstract: UAVs, also known as drones, are remotely operated or can be fully automated [1]. As the technology of computer vision and UAVs continue to evolve, it will not be absurd to suggest that a fully automated UAV will be able to do simple tasks such as picking up food at a restaurant and delivering it. However, this is only a minor task in what could be accomplished due to the evolution of technology. In the future, UAVs can be used as a platform to perform rescue operations, sending medical supplies and product delivery in obscure places. Since UAVs can be fully automated, it is for little to no margin for error. The project aims to use deep learning techniques to implement a working model that can accurately detect a specific landing zone in harsh and extreme conditions. The object detection model used needs to be quick to detect and be as accurate as possible. For this experiment, the model chosen is the SSD model for its speed due to this experiment being very time-critical, it is crucial to ensure that the object detection can be operated in a real-time environment. In conclusion, object detection for the landing zone of the UAV is possible depending on the use case. The pre-trained models which are available at the time of writing do not have a good balance between speed and accuracy. Choosing to prioritize one will leave the other undesirable.
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:SCSE Student Reports (FYP/IA/PA/PI)

Files in This Item:
File Description SizeFormat 
  Restricted Access
Final Year Project Report2.43 MBAdobe PDFView/Open

Page view(s)

Updated on Feb 1, 2023

Download(s) 50

Updated on Feb 1, 2023

Google ScholarTM


Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.