Road network restoration and emergency relief supplies delivery are two important aspects in the early stages of post-disaster emergency response. The main focus is on developing a joint scheduling plan for repair crews and transportation teams, so as to quickly connect life rescue routes and ensure that rescue personnel, emergency relief supplies, and equipments can be delivered to various demand points in the disaster area in a timely manner. However, most existing studies have considered road network restoration and emergency relief supplies delivery separately, making it difficult to meet the actual needs of emergency rescue and disposal. In this work, a road network model for joint scheduling of repair crews and transportation teams is first constructed. Next, the Markov decision process is adopted to simulate the activities of repair crews and transportation teams, in which the corresponding state spaces, action spaces, and reward functions are designed, respectively. Then, a joint scheduling algorithm for road network restoration and emergency relief supplies delivery is developed on the basis a customised double-layer interactive Q-learning. Finally, comparative experiments demonstrate that the proposed algorithm can improve the efficiency and effectiveness of road network restoration and emergency relief supplies delivery, and provide timely and reliable emergency relief supplies support for the rescue and disposal in the early stages of post-disaster emergency response.