Abstract: The architecture and surgical operation process of a brain surgery robot system based on vision localization technology is expounded to discuss in depth the mapping relations among four coordinates. Computer vision, robot and 3D-medical imaging reconstructing technology are combined together to help surgeons to complete brain's surgical operation. Marker registration method is adopted in robot system with vision localization instead of passive machine arm localization so as to improve greatly the exactness in operation as well as the intellectualization level. The mapping relations among each and every coordinate are thus studied further to work out the calibration of surgical operation actions and relevant planning. As a result, the robot realizes the planning posture. A theoretical foundation is therefore provided for practice.