Markerless Visual Servo Control of a Micro Helicopter with a Wireless Camera

摘要

In this paper, the authors present a markerless vision-based estimation and control algorithm for a micro helicopter with a wireless camera. The camera looks downward, and it captures an image of the ground at each time step. The proposed algorithm consists of vision-based estimation and switched control. The 3D location and posture of the helicopter are estimated by using template matching with automatic reference generation. The proposed algorithm also detects noisy images that are sometimes captured by the camera. The control signals are produced by a set of PID controllers if the current captured image is noiseless, and otherwise by a set of integral controllers. Long-time hovering is achieved over a high contrast and planar ground. No specified markers or prior texture information are required.