
2nd place internationally in the ERC drone task with 295/300 points, behind EPFL Xplore and tied with Brno Mars Rover: three autonomous flights, 8/9 sample tubes localized within 10 cm, and autonomous landing. I designed and implemented the onboard autonomy, real-time CNN perception and MAVLink mission software for 4Space UniOvi’s first competitive drone entry.
- 2nd place internationally in the ERC drone task with 295/300 points, behind EPFL Xplore and tied with Brno Mars Rover.
- Three consecutive autonomous flights over the drone Mars Yard.
- 8/9 sample tubes detected and localized with ground-position accuracy below 10 cm.
- Autonomous landing on the landing base prepared for the rover.
- Second consecutive in-person final in Kraków; 124 registered teams from 29 countries and 25 teams listed in final scoring.
- Designed and implemented the onboard autonomy, real-time CNN perception and MAVLink mission software for our first competitive UAV entry.
- Trained and optimized the CNN for onboard Raspberry Pi 5 frame processing.
- Used Optical Flow + LiDAR for GPS-free flight support.
- Validated the autonomous mission flow in Gazebo + ArduPilot SITL before real flights.







