We have developed flight hardware and software in-house, instead of adopting the common open platforms like ardupilot or pixhawk. This allows us to implement any guidance, control, and estimation algorithm that we develop for test and validation.
Indoor flight experiment
Our facilities include
- 12 VICON Valkyrie motion capture cameras in a netted indoor space of 40 ft by 20 ft
- Outdoor netted space of 20 ft by 20 ft
- Flight hardware based on NVidia Jetson and Orin computing module
- RTK GPS-based localization
- Flight software written in C++, python, and ROS
- GPU server powered by two NVidia A100 80G and one NVidia A100 40G
Selected flight software packages are shared as open-source software library at our github. Especially, the follwoing repositories have been popular:
- Geometric control of UAV (https://github.com/fdcl-gwu/uav_geometric_control)
- Python-Gazebo simulation of UAV (https://github.com/fdcl-gwu/uav_simulator)
Videos
Modular reinforcement learning
Autonomous landing on ship
Autonomous payload transportation
Quadrotor backflip in wind
Quadrotor backflip
Quadrotor hovering in wind
Delayed RTK GPS
Laser-guided landing
Autonomous aerial exploration
Autonomous ground exploration
Autonomous aerial transportation
Constrained attitude control