Aiming at the difficulty of accessing vehicles in underground garages for large buildings such as shopping malls, a guidance system for vehicle entry and exit is proposed in this paper by improved ant colony algorithm based on angle migration function and distance function. According to the actual demand, the system takes the free parking spaces, elevator positions and the number of curves of the guiding route as the optimization target, and plans the parking path closest to the elevator opening and the least number of turns under the premise of the shortest distance. Compared with the traditional shortest distance path planning algorithms, it has the characteristics of simplified path and higher efficiency, which is more in line with the actual needs of garage management.