Vision-based autonomous landing and nighttime trajectory estimation of quadrotors
Yang Liu · 2019
The agility and aerial view of drones can bring huge benefits to military, industrial and commercial fields. In recent years, huge improvements have been made in drone navigation technology. However, little effort has been put in drone navigation at night. This thesis is a cornerstone towards autonomous navigation for quadrotors at night. In the first part of the thesis, an vision-based autonomous landing program is implemented. It allows drones to land on a movable charging platform for longer battery life. This process provides a reliable solution with small error. It can also be transferred from grayscale camera to thermal camera in our future research, making long term autonomous task possible under poor light conditions. In the second part of this thesis, an optical flow based trajectory estimation program is implemented on thermal camera. This method can stablize a drone's position with more accuracy than GPS can in dark environment--Author's abstract