Learning to Rest: A Q-Learning Approach to Flying Base Station Trajectory Design with Landing Spots | Synapse