Feasibility of autonomous navigation above the Amazon forest by LiDAR images

This paper explores the feasibility of autonomous navigation above the Amazon forest without aid of a GNSS. The main difficulty of applying established techniques is the lack of features to find and match in an otherwise seamless sea of green. To solve this problem we propose using LiDAR cloud point data to estimate the terrain bellow the trees. We propose a simple algorithm and we show that, with a reasonable choice of resolution, it is possible to estimate the terrain and generate contour lines that provide the needed features to match with established feature matching algorithms.