Agile Autonomy: Learning High-Speed Flight (Ph.D. Thesis Defense Antonio Loquercio)

To date, only expert human pilots have been able to fully exploit the capabilities of quadrotors. Autonomous operation with onboard sensing and computation has been limited to low speeds. Indeed, agile quadrotor flight in unstructured environments requires low latency, robustness to perception disturbances, e.g. motion blur, and high precision, since the slack to avoid a collision is extremely limited. These characteristics push the boundaries of current state-of-the-art perception and control systems. The research of Antonio Loquercio shows that directly mapping sensory observations to navigation commands is a key principle to enable high-speed navigation. By mastering such a challenging navigation task, this work presents compelling evidence that the coupling of sensing and control is a fundamental step towards the development of a general-purpose robot autonomy in the physical world. Reference: Antonio Loquercio Agile Autonomy: Learning High-Speed Flight Antonio's webpage (publications, source code, slides): https://antonilo.github.io/ Google Scholar: https://scholar.google.com/citations?... Our research page on high-speed flight: http://rpg.ifi.uzh.ch/research_mav.html Our research page on drone racing: http://rpg.ifi.uzh.ch/research_drone_... For event-camera datasets, see here: http://rpg.ifi.uzh.ch/davis_data.html​​ and here: https://github.com/uzh-rpg/event-base... Affiliation: Antonio Loquercio is with the Robotics and Perception Group, Dept. of Informatics, University of Zurich, and Dept. of Neuroinformatics, University of Zurich and ETH Zurich, Switzerland http://rpg.ifi.uzh.ch/​​