Study and development of a reliable fiducials-based localization system for multicopter UAVs flying indoor

The recent evolution of technology in automation, agriculture, IoT, and aerospace fields has created a growing demand for mobile robots capable of autonomous operation and movement to accomplish various tasks. Aerial platforms are expected to play a central role in the future due to their versatility and swift intervention capabilities. However, the effective utilization of these platforms faces a significant challenge due to localization, which is a vital aspect for their interaction with the surrounding environment. While GNSS localization systems have established themselves as reliable solutions for open-space scenarios, the same approach is not viable for indoor settings, where localization remains an open problem as it is witnessed by the lack of extensive literature on the topic. In this thesis, we address this challenge by proposing a dependable solution for small multi-rotor UAVs using a Visual Inertial Odometry localization system. Our KF-based localization system reconstructs the pose by fusing data from onboard sensors. The primary source of information stems from the recognition of AprilTags fiducial markers, strategically placed in known positions to form a “map”. Building upon prior research and thesis work conducted at our university, we extend and enhance this system. We begin with a concise introduction, followed by a justification of our chosen strategies based on the current state of the art. We provide an overview of the key theoretical, mathematical, and technical aspects that support our work. These concepts are fundamental to the design of innovative strategies that address challenges such as data fusion from different AprilTag recognition and the elimination of misleading measurements. To validate our algorithms and their implementation, we conduct experimental tests using two distinct platforms by using localization accuracy and computational complexity as performance indices to demonstrate the practical viability of our proposed system. By tackling the critical issue of indoor localization for aerial platforms, this thesis tries to give some contribution to the advancement of robotics technology, opening avenues for enhanced autonomy and efficiency across various domains.