Aiming at the problems of low efficiency and poor accuracy of the existing picking robot, this paper proposes a novel optimization design method based on MBES and lean thinking. Firstly, the optimization analysis is accomplished by deeply excavating the potential lean factors in robot design requirements. Then the functional architecture is established by operational analysis and system analysis. Based on the analysis results, the lean optimization design of the robot is studied. Specifically, by constructing the value flow of the picking process, the picking manipulator is optimized from a fully flexible arm to a rigid-flexible hybrid structure. Furthermore, in order to reduce the waste of time in the picking process, the DMAIC method is used to optimize the action of the picking equipment on both sides to operate separately, so as to improve the picking efficiency.