DDPG-based Deep Reinforcement Learning for Loitering Munition Mobility Control: Algorithm Design and Visualization

Abstract

Drone technology is estimated for its potential to be applied in many industries, including logistics, broadcasting, telecommunications, and warfare technology. In particular, in the field of modern warfare such as the current war in Ukraine, the use of drones has become an essential element. This paper includes a loitering munition to attack a single ground target in the scenario. A simulation environment for drone attack is built based on the 3D platform Unity, and learning is performed by applying DDPG, a reinforcement learning algorithm that can be used in continuous action space. Through the specific result, it is possible to achieve our purpose to attack target exactly.