Simulation · Research infrastructure

Testing drones in simulation

At UAH, I built simulation environments and the tooling around them, from multi-aircraft missions to a Unity environment for human–AI research.

Organization: University of Alabama in Huntsville
My contribution: Simulation environments and test infrastructure

Illustration of three aircraft searching a mission area with tents, mannequins and marked caches. — An illustration of the search scenario, not a recording of a Gazebo test run.

Simulating a mission

In Gazebo, we ran whole missions for SAPIENCE with PX4 software in the loop. That included custom airframes and cameras, as well as the mannequins and marked caches the aircraft had to find.

I built distributed simulation support, the CI/CD and container workflows used to run these tests, and the contract tests and fuzzers that exercised the running services. A grader scored each mission from its logs.

Running a simulated fleet

I ran 25+ drones in PX4 SITL on consumer desktop hardware using K3s. In the field, we deployed three real aircraft the same way, with a smaller set of services in the flight configuration.

Try the mission demo How deployment worked

Human–AI research in Unity

In a separate project, I built a Unity search-and-rescue environment with variable terrain, obstacles, and interactive mission conditions. It was used in more than 100 experiments.

You can find the related work in my publications.