Design and Implementation of a DQN Based AAV

Abstract

The Deep Q-Network (DQN) is a method of deep reinforcement learning algorithm. DQN is a deep neural network structure used for the estimation of Q value of the Q-learning technique. The authors have previously developed a simulation system on DQN-based behavioral control methods for actuator nodes in Wireless Sensor Actor Networks (WSANs). In this paper, an Autonomous Aerial Vehicle (AAV) testbed is designed and implemented for DQN-based mobility control. We evaluate the performance of the AAV testbed for a indoor single-path environment. For simulation results show that the DQN can control the AAV.