Abstract: It can improve the fuel economy of vehicles when they pass through traffic light intersection to acquire traffic light intersection information in front by a high-precision map and possess the ability of automatic driving to exceed human's global decision-making and planning. In this study, a transient fuel consumption model and dynamic programming algorithm were used to calculate the vehicle fuel consumption and corresponding vehicle speed sequence, taken as the edge set weights of the traffic light spatio-temporal topology map. And then an A-star algorithm was used to plan the vehicle speed trajectory in order to improve the vehicle fuel economy. The results show that when the speed trajectory planned in this paper passes through four traffic light intersections, it can improve fuel economy by 33.6% and reduce the travel time by 17.7% at most compared with the most economical speed of flat roads.