Autonomous drone racing is gaining a great popularity among researchers. Recently, a number of projects have been launched to encourage rapid advances in the field, e.g. Fast Lightweight Autonomy (FLA), European Research Council’s AgileFlight, and the AutoAssess project. To achieve high performance, racing drones require real-time algorithms that are robust to motion blur, high dynamic range, aerodynamic disturbances, and appearance of opponents. ETH Zurich reached an important milestone when its RL policy outraced all human pilots. The project will focus on the development on Russian first vision-based, autonomous drone racing with leveraging Reinforcement Learning (RL) and Imitation Learning (IL). The experiments in drone racing arena with gates with sub-millimeter precision mocap system will be conducted.

Article on the project.

V. Serpiva, A. Fedoseev, S. Karaf, A. A. Abdulkarim, D. Tsetserukou, “OmniRace: 6D Hand Pose Estimation for Intuitive Guidance of Racing Drone,” in Proc. IEEE/RSJ Int. Conf. on Intelligent Robots and Systems, (IROS 2024), Abu Dhabi, UAE, 14-18 October 2024, 2024, in print. (No. 1 Conference in Intelligent Robotics and No. 2 in Robotics, Core2023 A, Scopus and WoS, H(SJR)=150). Arxiv: https://arxiv.org/abs/2407.09841

Project status: under development.