A Survey of Deep Learning Techniques and Computer Vision in Robotic and Drone with Applications

¹ National School of Electronics and Telecommunications, Information Technology, University of Sfax, Tunisia
² M. SC Information Technology, University of Babylon, Babil, Iraq
³ Professor in Higher Institute of Biotechnologies of Sfax, ISBS, University of Sfax, Tunisia

^* Corresponding author: basic.maysoon.maroof@uobabylon.edu.iq

Abstract

The methods of deep learning have lately demonstrated outstanding outcomes of robotic objects such as imagination, localization and striping. Its exceptional abilities in order to learn idealizations from complicated data gathered in the real world ambiance make it perfect for a high range of independent applications of robot. Simultaneously, unmanned aerial vehicles are becoming more used for a variety of civilian stints ranging from security, superintending, and disaster relief, extraditing of package and repository arrangement. A thorough exegesis one of the primary deep learning techniques is also supplied. A set of the main difficulties in using deep learning with UAV-based solutions. Even still, autonomous navigation remains a challenge where computer vision technologies can shine. As a result, development the forecast made by the network and the ground-truth attention distribution, increased the use of vision systems and algorithms, been a major focus of studies conducted recently. An organized mapping investigation is conducted to gain a broad perception of subject. Some studies provide a thorough examination of addressing computer vision in relation to the following independent unmanned aerial vehicles vision establish chores such as navigation, control, back trace and sense.