Deep Reinforcement Learning for automatic landing of airships on Ahagar terrains

Abstract

Algeria is a vast country holding many touristic areas that need to be discovered. The problem arises from the difficulties of landing an airship in some of these areas due to their hard nature. In this paper, we discuss, for the first time, the problem of automatically landing an airship on dangerous terrain in the Ahagar environment. Due to its effectiveness, deep Q-learning (DQL) has been employed for realizing such a task. Our proposed landing model has yielded satisfactory results in normal cases. To examine the stability of the proposed model, it has been subjected to two random forces which are wind and engine failure. The proposed model has proven stability to a certain extent after which the landing becomes dangerous. The proposed model can be employed for two tasks, the first one is the automatic landing of airships on Ahagar, and the second one is the prediction of landing outcomes in case of the presence of random forces.