Image Segmentation-Based Unmanned Aerial Vehicle Safe Navigation

This work proposes a vision-based guidance scheme for an unmanned aerial vehicle navigating through urban environments while seeking a predefined goal point. Optical flow of image corner feature points is considered to segment obstacles from the image. An obstacle-avoidance guidance law is proposed to avoid the segmented obstacles. Additionally, detecting open space between segmented obstacles, a passage-following guidance law also presented for intelligent decision making. Analytic comparison with an existing methodology is carried out to highlight superior obstacle-avoidance properties of the proposed strategy. Simulations are carried out in a three-dimensional environment for single and multiple obstacles. Results comply with the analytic findings and present a much improved avoidance performance as compared to existing optical flow-based methods.