This paper investigates the implementation and flight test of realtime vision guided autopilot system based on virtual instrumentation platform. A graphical design process via virtual instrumentation platform is fully used for the image processing, communication between systems, vehicle dynamics control, and vision coupled guidance algorithms. A significatnt ojective of the algorithm is to achieve an environment robust autopilot despite wind and an irregular image acquisition condition. For a robust vision guided path tracking and hovering performance, the flight path guidance logic is combined in a multi conditional basis with the position estimation algorithm coupled with the vehicle attitude dynamics. An onboard flight test equipped with the developed realtime vision guided autopilot system is done using the rotary UAV system with full attitude control capability. Outdoor flight test demonstrated that the designed vision guided autopilot system succeeded in UAV's hovering on top of ground target within about several meters under geenral windy environment.