Design and implementation of homography-based vision-aided inertial navigation of UAVs

This paper presents a novel vision-aided inertial navigation system for navigating a small-scale unmanned aerial vehicle (UAV) in unmapped and GPS-denied environments. Suppose the UAV is equipped with a low-cost inertial measurement unit (IMU), a downward-looking camera, a barometer and a compass. By fusing the measurements of the four sensors using an extended Kalman filter, we can obtain drift-free estimates of the velocity and attitude of the UAV. In the meantime, the drift of the UAV position can be significantly reduced and the unknown biases in the IMU measurements can be identified online. The vision measurement in our navigation system is a homography matrix, which can provide the velocity and attitude information of the UAV. We show flight experimental results to verify the effectiveness and robustness of the proposed navigation system.