Mothership delivery system using reinforcement learning

Abstract

We address drone delivery system with mothership which is advanced version of vehicle routing problem. The mothership as moving depot replaces static depot, drones replacing trucks to improve the transportation efficiency. In this paper, methods to solve this problem is proposed and verified with numerical experiments. The main contributions of this paper are 1) the advanced problem is suggested and conditions to improve the optimization level of classical problem. 2) A environment for reinforcement learning is defined to solve the problem and various tests to prove that has no fault. 3) The reinforcement learning algorithms are proposed to solve the problem and represents superiority by comparing other possible algorithms.