Existing research on airport ground support vehicle scheduling mostly focuses on single-type vehicles. Challenges such as task duration uncertainty and temporal constraints are faced by multi-type vehicle collaborative scheduling. A robust collaborative scheduling optimization method considering task duration distribution is proposed. Task duration probability distribution models (TDDM) were established based on historical operation data. A collaborative scheduling optimization model for multi-type vehicles was constructed. Task duration, task priority, and inter-task temporal constraints were considered. Total flight turnaround delay time and total vehicle travel distance were minimized as optimization objectives. A dual-population genetic algorithm (DPGA) was designed to solve the model. Parallel forward and backward searches were performed. Time buffers were embedded into the scheduling scheme to absorb task duration fluctuations. Simulation validation experiments were conducted using actual airport operation data. Pre-scheduling schemes were generated using expected task duration values. Robustness was evaluated via Monte Carlo sampling. The average flight turnaround delay time is reduced by 10.0% compared with traditional deterministic scheduling methods. Good stability is demonstrated under different perturbation intensities. Decision support is provided for dynamic resource optimization in uncertain environments.