Towards Autonomous Landing of a Quadrotor using Monocular SLAM Techniques

Use of Unmanned Aerial Vehicles have seen enormous growth in recent years due to the advances in related scientific and technological fields. This fact combined with decreasing costs of using UAVs enables their use in new application areas. Many of these areas are suitable for miniature scale UAVs Micro Air Vehicles (MAV) which have the added advantage of portability and ease of deployment. One of the main functionalities necessary for successful MAV deployment in real-world applications is autonomous landing. Landing puts particularly high requirements on positioning accuracy, especially in indoor confined environments where the common global positioning technology is unavailable. For that reason using an additional sensor, such as a camera, is beneficial. In this thesis, a set of technologies for achieving autonomous landing is developed and evaluated. In particular, state estimation based on monocular vision SLAM techniques is fused with data from onboard sensors. This is then used as the basis for nonlinear adaptive control as well trajectory generation for a simple landing procedure. These components are connected using a new proposed framework for robotic development. The proposed system has been fully implemented and tested in a simulated environment and validated using recorded data. Basic autonomous landing was performed in simulation and the result suggests that the proposed system is a viable solution for achieving a fully autonomous landing of a quadrotor.