Abstract:  In order to aquire accurate and robust positioning results, a low cost monocular camera and an IMU(Inertial Measurement Unit) are combined by sensor fusion for the ego-positioning of MAVs (Micro Air Vehicles). In the visual positioning module, three of the most representative state-of-the-art monocular visual positioning approaches, which are originally designed for vehicles, are evaluated in different conditions. Based on the experiment results, the pros and cons of each method are analyzed and discussed. Considering the limitations of vision-2only approaches, an EKF (Extended Kalman Filter) is used to fuse the visual positioning with an inertial sensor based on a loosely-coupled framework, to compensate for the instability of vision. The experiment results demonstrate that the visual-inertial sensor fusion can effectively reduce the positioning errors in unstable visual cases.

Key words: monocular vision, sensor fusion, inertial measurement, micro air vehicles(MAVs),  positioning