In this article, a new deep reinforcement learning (RL) method, called asynchronous advantage actor-critic (A3C) method, is developed to solve the optimal control problem of elevator group control systems (EGCSs). The main ...