Implementation of a Visual SLAM System

Abstract

The goal of this project is to develop a system to compute the position and orientation of an UAV using a monocular camera. To achieve that, the WOLF library will be used. WOLF is a library thought to solve generalized simultaneous localization and mapping (SLAM) and visual odometry problems. Derived classes will implement the algorithms needed to track features from the images obtained through the sensors. Constraints between the features and other features, or with landmarks, will be created to form a factor graph. An external solver will iterate to find the optimal state, by minimizing the cost associated to all the constraints. Ideally, our system can be used together with an inertial model measuring rotational velocities and translational accelerations, and tested with an unmanned aerial vehicle (UAV) in simulation and in real.environments.