This thesis aims at solving scheduling problems in uncertain environment. The goal is to develop algorithms which optimize order preparation in warehouses. Several constraints have to be considered such as available workforce, physical space available for dealing simultaneously with several orders, breaks in workforce plannings, ... Moreover not all the jobs are known before the start of the algorithm and some more jobs are revealed during the execution of the schedules generated beforehand. Hence developed algorithms must adapt calculated solutions in order to take into account those new jobs. Furthermore, historical data allows us to have stochastic knowledge about the jobs which will be revealed. Such data are used in order to improve algorithms performances.