Lost in gates: enhancing state estimation in high speed autonomous drone racing

This thesis delves into the integration of Artificial Intelligence (AI) within the realm of autonomous racing drones. Traditionally, the racing industry has led the way in advancing resilient and streamlined systems, with recent emphasis shifting towards the implementation of autonomous driving mechanisms. The intricacies of this field present complex challenges, demanding the development of precise control and perception systems that operate with minimal reaction times and constrained resources. The research, conducted within the Drone Racing team at the Autonomous Robotics Research Center of the Technology Innovation Institute, primarily focuses on advancing perception and state estimation systems for autonomous racing drones. Central to the study is the introduction of a novel high-speed autonomous drone racing multimodal dataset and an innovative map-based, perceptually aware state estimation technique. This work is instrumental in pushing the boundaries of autonomous drone racing technology, offering valuable insights and solutions that contribute to the broader advancements within the field of autonomous systems.