At present, a lot of airline luggage relies on manual stacking, and the existing 3D automatic packing stacking algorithm still has the problem of low space utilization for strong heterogeneous luggage packing, which urgently needs algorithm improvement research. To this end, mathematical modeling was conducted based on the actual needs of airport luggage check-in and stacking. A certain amount of information such as checked luggage size, weight, and loading sequence was obtained in advance through front sensors, and the stacking posture of the luggage was optimized and adjusted. Then, a fitness function was designed and combined with the heuristic search principle of remaining space to improve the original Artificial Hummingbird Algorithm (AHA), forming a more effective packing strategy. The proposed algorithm was used to simulate the stacking of 3×200 pieces of real airport luggage. The simulation results showed that the algorithm could quickly complete the stacking calculation and improve the average space utilization of the luggage cart by 9.3%. The improved algorithm has short computation time, high space utilization, and strong packing stability, which can provide strong support for the modernization and intelligence of airport logistics.