Abstract: In order to improve the stiffness performance of the robot and reduce the milling processing errors, the stiffness of a 6 DoF (degree of freedom) robot equipped with a milling actuator is optimized. Firstly, a robot stiffness mapping model is established using the virtual work principle. Secondly, the joint stiffness is obtained by the designed identification experiments. Then, the optimized pose of the robot is solved using the genetic algorithm, taking the isotropy of the milling force elliptic plane as the optimization index. Finally, the overall stiffness of the robot before and after the pose optimization are compared and analyzed. And a robot milling test is performed to verify the effectiveness of the pose optimization. Consequently, the milling flatness can be improved by 45%. The optimization method can guide the milling processing tasks of the serial industrial robot on the large spacecraft cabin, and improve the processing quality.