Training in “First Person View” Systems for Racing Drones

This paper presents the development of a two-degree-of-freedom camera mount, equipped with a head motion tracker, designed to improve the training process for drone racing. Our research has resulted in three main contributions: We have implemented an algorithm that is capable of detecting race gates in real time, thus allowing the operator to quickly identify and approach these gates in a more controlled manner. We have incorporated a two-axis camera movement function, controlled by a head motion tracker, which provides the operator with a superior sense of spatial orientation in free flight while the use of a single degree of freedom helped in racing drone trainings. We developed a structure light enough to ensure minimal impact on the drone's center of gravity. These innovations, reinforced by artificial intelligence, are intended to improve the operator's performance and overall experience, helping during the training process for drone racing as well as in free flight. The goal of our study is to contribute to ongoing efforts to improve human interaction and control in UAV operations, with the ultimate goal of facilitating a safer, more efficient and enriching flight experience. Finally, the NASA Task Load Index questionnaire was used to assess whether the head tracker effectively assists the end user.