In order to study the impact of future autonomous vehicles on the road network capacity, the vehicles on the road network were divided into two types: autonomous vehicles and ordinary vehicles. According to the different path selection behaviors of the two types of vehicles, a bi-level programming model of the road network capacity was constructed considering the influence of driverless vehicles. The upper layer model was the maximum demand under the condition of road capacity constraint. The traffic demand between different OD used a uniform growth multiplier. The lower model was a mixed routing behavior model considering the influence of the autonomous vehicles. Among them, traditional vehicles aimed at minimizing individual travel cost. Autonomous vehicles pursued the optimal systems goal. The heuristic algorithm for the bi-level programming was designed based on the idea of iterative balance. The validity and feasibility of the model and algorithm were verified by a simple urban road network. The result indicates that:①When the market penetration of the autonomous vehicles is low, the autonomous vehicles have little effect on the path selection of the ordinary vehicles. The fluctuation of the road network of the mixed equilibrium flow is small, and the road network capacity increase is not obvious. ②As the market penetration rate increasing, the road network capacity firstly shows a slowly enhancing state. The autonomous vehicles will take over while reaching a certain proportion. The total numbers of available routes are increasing. Therefore, the traffic volumes are more evenly distributed on different road segments. The road network capacity growth trend increases rapidly. The whole system is in the optimal state and the road network capacity is maximized if only autonomous vehicles exist in the network.③Although the total network impedance increased due to the increase of the road network capacity, the average travel cost of the network traveler does not change too much.