This thesis starts with an introduction to Simulataneous Localization and Mapping (SLAM) and more background on Visual SLAM (VSLAM). The goal of VSLAM is to map the world with a camera, and at the same time localize the camera in that world. One important step is to optimize the acquired map, which can be done in several different ways. In this thesis, two state-of-the-art optimization algorithms are identified and compared, namely the g2o package and the SBA package. The results show that SBA is better on smaller datasets, and g2o on larger. It is also discovered that there is an error in the implementation of the pinhole camera model in the SBA package.

For a long time stereo-cameras have been deployed in visual Simultaneous Localization And Mapping (SLAM) systems to gain 3D information. Even though stereo-cameras show good performance, the main disadvantage is the complex and expensive hardware setup it requires, which limits the use of the system. A simpler and cheaper alternative are monocular cameras, however monocular images lack the important depth information. Recent works have shown that having access to depth maps in monocular SLAM system is beneficial since they can be used to improve the 3D reconstruction. This work proposes a deep neural network that predicts dense high-resolution depth maps from monocular RGB images by casting the problem as a supervised regression task. The network architecture follows an encoder-decoder structure in which multi-scale information is captured and skip-connections are used to recover details. The network is trained and evaluated on the KITTI dataset achieving results comparable to state-of-the-art methods. With further development, this network shows good potential to be incorporated in a monocular SLAM system to improve the 3D reconstruction.