The scarcity of labor resources and the sustained increase in human capital costs have escalated human risks and labor costs in labor-intensive enterprises. This has been mitigated by the advent of mobile robot fulfillment systems (RMFSs), which have reduced the high labor intensity in warehouse picking systems and the workloads of picking personnel. The picking task in RMFSs is executed collaboratively between mobile robots and picking personnel. The behavior and working status of mobile robots are precisely regulated by the system, resulting in consistent and reliable performance regarding their work status and efficiency. Nevertheless, the efficiency of picking personnel remains contingent on various factors, resulting in uncertainties in the overall system efficiency. Hence, it is crucial to prioritize ensuring and enhancing the efficiency of the picking workstation during the overall planning of an RMFS. In an RMFS, the most subjective and uncontrollable agent is the picker who executes the picking operation at the picking workstation. When there is an adequate number of robots working continuously, the efficiency of pickers cannot be constantly maintained. If the fatigue of pickers is not considered when developing a robot scheduling scheme, the picking workstation will become a system bottleneck. This paper presents our investigation of the effect of fatigue factors on picking efficiency and proposes a human-robot collaborative robot scheduling scheme that considers picking fatigue.The continuous work required of picking personnel inevitably leads to fatigue. In a multi-robot scheduling scheme that fails to account for picking personnel fatigue, there will be a significant and random deviation between the actual and expected picking completion time. As the number of tasks completed increases, this deviation will increase. In this study, we studied the effect of picker fatigue factors on robot scheduling decisions by comprehensively analyzing the causes and effects of picking fatigue on picking efficiency. Based on our insights, we propose multiple objectives.A robot-scheduling model was established that considered picker fatigue and aimed to minimize the task completion time of the picker. In this model, the working time of the picker is fully utilized. However, multiple robots queuing in the picking workspace may result in conflicts, blockages, and increased robot running costs. Subsequently, another model was established to minimize the queuing time of all robots, which balanced the multiple robots queuing for a long time solely to achieve picker efficiency. Finally, a multi-objective multi-robot scheduling model was established to minimize picker task completion time and queuing time for all robots, with the results of the robot task allocation serving as decision variables. Taking into account the psychological perception pressure caused by the continuous queuing of a large number of robots that can reduce picker efficiency, we propose to use a constrained approach to handle picker task completion time and optimize the queuing time of all robots to ensure that picker efficiency remains within the expected range of the manager. A genetic algorithm is employed to solve this multi-objective model.Finally, simulation experiments were conducted. The results revealed that the multi-objective model considering picker fatigue factors proposes a human-robot collaborative multi-robot scheduling scheme that enhances the picker workstation efficiency and allows the system to operate efficiently, smoothly, and stably. The greater the picking complexity of an order, the more pronounced the optimization effectiveness of the proposed multi-objective model. For the same batch of orders, prioritizing allocating low-difficulty picking tasks to robots results in slowly accumulated fatigue in pickers and a higher picking efficiency. When allocating goods of different picking difficulty levels to the same rack, efforts should be made to balance the picking difficulty differences between different rack levels. Picking efficiency improves when fatigue accumulates slowly for pickers during picking operations. The model demonstrates adaptability to different picking personnel and various types of stored products by analyzing parameters influencing picker fatigue. This aligns with the organizational management philosophy based on organizational initiative theory and enables picking personnel to maintain high physical and mental immersion in the task.By optimizing robot scheduling to enhance the effectiveness of robot-assisted picking, it is possible to enhance the subjective initiatives of the pickers and make the RMFS highly efficient operationally, providing theoretical guidance for enterprise management decision-making.