Position control of a quadrotor UAV using stereo computer vision

Abstract

Quadrotors are a recent popular consumer product, prompting further developments with regards to combining attitude and position control for such unmanned aerial vehicles (UAVs). The quadrotor’s attitude (orientation) can be controlled using measurements from small and inexpensive sensors such as gyroscopes, accelerometers, and magnetometers combined in a complementary filter. The quadrotor’s position can be controlled using a global positioning system (GPS); this has shown to be an effective method but is not reliable in urban or indoor environments where quadrotors may be flown. An alternative method such as visual servoing may provide reliable position control in urban environments. Visual servoing uses one or more on-board cameras to detect visual features in their field-of-view. If more than one camera is used, a more accurate position can be determined. This work discusses the the application of stereo visual servoing to a quadrotor for position control combined in a Kalman filter with IMU readings to provide a better estimate of its position.