In order to consummate the master-slave robot system used in vascular interventional procedures, a closed loop of force is rebuilt. Two micro force sensors are integrated with a catheter, which can measure those forces between the tip or side wall of the catheter and vascular walls respectively at the same time. Given that the master-slave mode is different from the traditional hands-on operation and that it is necessary to ensure the sustained effect of the valuable experience from surgeons, a two-input one-output model in fuzzy inference is adopted to accomplish the fusion of the two sensors' data. By means of calibrating membership functions and establishing a table of fusion rules, experience is converted into numerical strategies about the force feedback, which not only improves the force reflection under low sampling rate, but also characterizes collisions accurately with different resolutions. It ensures the safety, applicability and efficiency of the robot system.