Optimal scheduling of the electric vehicle charging process is beneficial to the safe and stable operation of power grids. It improves road traffic efficiency, facilitates renewable energy utilization, and reduces the charging time and costs for users. Deep reinforcement learning can effectively solve the problems caused by different randomness and uncertainty in the optimal charging scheduling. This paper summarizes the working principle of deep reinforcement learning first, and makes the comparison of the characteristics and applications among different types of reinforcement learning. Then, the research results of deep reinforcement learning for EV charging scheduling are summarized in terms of both static and dynamic charging scheduling, and the shortcomings of existing research are analyzed. Finally, future research directions are discussed. This work is supported by the National Natural Science Foundation of China (No. 62176088).