11. google map oriented visual navigation for uavs
Name: Mo Shan
Grad Year: 2021
Traditionally, UAV navigation is entrusted on GPS, which may be unreliable. We describe a framework to perform geo-referenced navigation based on Google Map. Localization is achieved via registering the on-board image to the reference, which relies on Particle Filter. HOG is used in a holistic manner to describe the images while Optical Flow is used for prediction and localisation with the absence of reliable match between the on-board image and the map. Our approach named HOP is able to handle multi-modal image registration in real time. Offline test using the flight data demonstrates the effectiveness and efficiency of our approach.