Path Planning and Obstacle Avoidance Based on Reinforcement Learning for UAV Application

Applications of drones in the military and daily life have increased in recent years. However, it is necessary to have obstacle avoidance capability. Path planning is also needed for automated tasks. In this study, path planning and obstacle avoidance based on a reinforcement learning algorithm are implemented in an unmanned aerial vehicle (UAV). For global path planning, a Q-learning algorithm is utilized to find the path between waypoints first. Then the local obstacle avoidance system is applied to train the UAV by the Deep Q-learning algorithm in Microsoft AirSim unreal environment.