This paper focuses on the caldron feeding process during the traditional distillation of Chinese liquor, and designs a robot operation system to accomplish this process by studying the typical human operations. Firstly, this paper proposes a generalized coding algorithm for the caldron feeding trajectory based on B-spline curve Gaussian model. The information of artificial movements is collected on the workplace and imported into the generalized model to obtain the trajectory characteristics of the artificial process. After that, the proposed method separates the working plane to 8 areas by analyzing the characteristics of standard trajectory, and designs a hopper mechanism with a controllable material flow rate. By considering of the structure of hopper mechanism and the workplace layout, the cartesian trajectories of the robot for each area are calculated. To make the process more efficient, this paper calculates the largest velocity of the robot by analyzing the dynamic model of the robot for each area with the limit of joint movement. Then the control strategy of feeding height is proposed which is influenced by the material flow rate and robot velocity. Finally, the robot operation system is confirmed to be more efficient by repeatedly caldron feeding tests.