Autonomous Quadcopter Navigation of Trails Using Convolutional Neural Networks

Abstract

This research focuses on developing an artificially intelligent quadcopter that implements computer vision through Deep Neural Networks to autonomously navigate trail paths for search and rescue. The challenge of search and rescue along trails has normally been a completely manual process, but following the recent growth in UAV technologies and big data processing, the potential of solving this problem autonomously has become feasible. Currently, a variety of startups and research groups have begun working towards developing similar technologies for different applications, but very few have actually demonstrated successful prototypes. Thus, there is great value in pursuing research and experimentation in this direction for both the safety of hikers around the world and the advancement of UAV technologies through machine learning.