Abstract: The scheduling problem of different types of robotic hand stacking system at the cold-end of float glass was studied. The problem was divided into two stages: cutting stage and grasping stage. In the cutting stage, a certain number of sheets were rationally arranged on the glass belt. In the grasping stage, High-Speed-Stacker(HSS) and Pick-on-the-Fly(POF) manipulators were used to stack sheets of different specifications, respectively, and the goal was to minimize the waste rate of glass. According to whether the two types of manipulators work at the same time, the problem proposed two solutions: mixed grasp and classified grasp, and the optimization method of Balance Unit. In the grasping stage, a combination variation cuckoo search algorithm (CVCS) was proposed, which introduced internal and external reverse order and multi-field insertion mutation operations to prevent the algorithm from falling into local optimum. In the simulation experiment, the response surface analysis method was used to calibrate the parameters of CVCS. CVCS algorithm was compared with MMGA, BPSO, DABC and CS algorithm. The experimental results show the superiority of CVCS algorithm in solving this problem.