optimal_quad_control_RL

Data underlying the publication: MonoRACE: Winning Champion-Level Drone Racing with Robust Monocular AI

This dataset supports the publication “MonoRACE: Winning Champion-Level Drone Racing with Robust Monocular AI”. The research investigates robust autonomous high-speed drone racing using a single forward-facing monocular camera and an onboard IMU, with the goal of enabling reliable gate detection, state estimation, and control during aggressive flight.

The dataset contains flight-log data from eight autonomous drone racing flights, providing experimental data collected during the development and evaluation of the MonoRACE system. The flight logs contain onboard sensor and flight-control data used to analyse the behaviour and robustness of the system during racing.

In addition to the flight data, the dataset includes the software components used in the experiments: the MonoRaceGate implementation for monocular gate-based navigation, the INDIFlight flight-control software, and the Optimal Quad Control reinforcement-learning controller developed for the A2RL 2025 autonomous drone racing competition. These components provide the research and implementation context needed to interpret and potentially reuse the data.

The dataset therefore combines experimental flight logs with source code for the perception, navigation, reinforcement-learning control, and low-level flight-control components used in the research.