Terrain-aware Low Altitude Path Planning

Abstract-- In this paper, we study the problem of generating low-altitude path plans for nap-of-the-earth (NOE) flight in real time with only RGB images from onboard cameras and the vehicle pose. We propose a novel training method that combines behavior cloning and self-supervised learning, where the self-supervision component allows the learned policy to refine the paths generated by the expert planner . Nap-of-the-earth (NOE) flight is an important tactic to reduce the exposure of an aircraft during flights. For a piloted aircraft flying at high speed, NOE flights are intensive as they require the pilots to extract terrain information and react very quickly to new information. Therefore, it would be beneficial to automate some of the tasks, such as path planning [1]. Moreover, solutions relying only on sensors that do not emit (e.g. Automating part of the navigation task for NOE flights has been studied in numerous previous works.