UC Santa Cruz

The ubiquity of Unoccupied Aerial Systems/Vehicles (UAS/V) in society hasspurred many different functional purposes for them. In many of these applications, including visual sensing is advantageous for precise control and situational awareness. Visual “Simultaneous Localization and Mapping”, or Visual SLAM, has come to the fore as an effective approach for proprioception in UAS. In this thesis, two explorations on the relationship between Visual SLAM and UAS are offered: (1) an in-depth analysis of the performance of the algorithm when flown in a sparse feature space onboard a multicopter platform, and (2) the design and simulation of an application example of Visual SLAM in a terrain-following modality of flight. Through these contributions, the viability and reliability of Visual SLAM is validated for practical use on small aerial vehicles.