To solve the cooperative task scheduling problem under uncertain conditions for agent group, an interval programming mathematical model was established from the aspect of improving robustness of task allocation scheme. A non-dominated sorting algorithm in interval pattern was proposed for solving this model. Since the model aimed to minimize task completion time while maximize task completion quality, dominance relationship between different individuals was defined. Then the algorithm guided the evolution of population by using random genetic operator and heuristics operator in coding space and removed chromosomes to keep the size of population by adopting circular crowded sorting strategy in decoding space. Experiment results show that the proposed method is feasible and effective and can obtain robust and superior task allocation schemes under uncertain conditions.