Visual and inertial odometry based on sensor fusion

Given conventional procedures of visual odometry based on stereo vision could have problems with detection if an environmental object is not in a fixed location. It is important to consider the possibility of adding more sensors. In this case, an IMU presents an additional information source. For this reason, using sensor fusion is important, it is a way to improve methods for pose estimation. This research was focused on developing a Visual Odometry (VO) method based on stereovision and Inertial Odometry (IO) by applying sensor fusion. The main goal of this method was to improve the possibilities of robot autonomous navigation. The pose estimation using this methods has been planned as an option for not losing accuracy with moving objects or light noise that affect the visual odometry. This project has been tested in a mobile platform, and designed for future drone applications.