The challenge

The client needed repeatable tests for a UAV ground control station, but access to physical hardware was a bottleneck.

Architecture and approach

We built a Python mock using pymavlink to handle the small subset of MAVLink communication needed to trigger events in QGroundControl and respond to requests in predefined ways. For these tests, it provided a lightweight alternative to a full simulation in Gazebo. QGroundControl ran in a Docker-based container environment, with Squish automating interactions with its graphical interface.

Running tests in CI

The mock and GUI tests formed a regression test setup that could run in CI without a physical drone. The project used GitLab CI alongside Docker, Python, pymavlink and Squish.

Outcome and scope

The resulting regression suite could run without a drone in the room. The mock covered only the communication needed for these tests; it does not replace validation on physical hardware.

Discuss a similar challenge

[email protected]

← Back to selected work