Simultaneous Localization and Mapping Using Stereoscopic Computer Vision for Autonomous Vehicles

In creating a robotic system capable of autonomously making decisions based on an environment, it must be capable of visualizing and recording its surroundings. This thesis focuses on the use of stereoscopic imaging from two offset camera feeds to create a 3D image. The robot is capable of combining multiple 3D images to create a 3D model of the entire area of operation. The robot must be able to generate a 3D image from the cameras and convert the 3D images into a digital map. Furthermore, the robot must be able to detect its location within the digital map for navigational and mapping purposes. This process of simultaneously localizing and mapping (SLAM) is a revolutionary procedure used to generate maps in real time. By combining these three aspects, the robotic system can generate an accurate 3D map for the region of operation