Documenting the development of autonomous flying systems. A ZHAW informatics team's journey through prototyping, flight experiments, and intelligent navigation.
The semester is over. The LionBee flies autonomously, the ELRS port is a genuine open-source contribution, and the MAVLink pipeline works. Here is an honest look at what the team built, what stayed unfinished, and what we would do differently.
The drone completed its first full autonomous mission outdoors — take off, fly to a waypoint, land. PID tuning is done and the hardware is flying stably. But GPS accuracy indoors is a real problem, and the simulation drift issue on the software side is still open.
GPS is stable, blackbox logging works, and the hardware is finally fully operational under iNav. But the software team is wrestling with persistent drift during simulation flights, and the hardware team has hit an architectural wall: MAVLink over SPI alongside ELRS has never been done before.
The software team is reading real sensor data via MAVLink — but mission planning is proving tricky. Meanwhile, the hardware team completed all seven steps of the ELRS firmware port, and the drone can now be controlled with a real radio transmitter under iNav.
The simulation now uses the real LionBee model with accurate weight and thrust. MSP control is fully working. And after a meeting with the project owners, the team has decided to port the SPI ELRS driver from Betaflight to iNav — a big challenge, but the right call.
With the development environments in place and the drone assembled, this week was about pushing further into the actual technical work. The software team took their first real steps controlling the drone through code, while the hardware team dug deeper into the board's port configuration trying to solve some stubborn hardware recognition issues.