The Wright Brothers Institute was tasked with researching the effectiveness of drone-based search and rescue. In order to measure this effectively with a repeatable situation, they needed a realistic simulation which uses the same drone control software as the real world.
Developers needed to construct a simulation for pilots to train with search-and-rescue UAVs, but the software had to be designed around the firmware and hardware of the drone. The simulation had to accurately represent the capabilities of the drone and simulate an environment that pilots would practice in.
Wright Scholars created a simulation in Unreal Engine 4, utilizing AirSim to simulate drone physics. In order to make the simulation as close to the real world situation as possible, they used Ardupilot firmware in WSL (Windows Subsystem for Linux) which linked to AirSim. They also used QGroundControl to control the drone, which made the simulation extremely similar to reality.
Students made a simulation with an AI target that maps and alters its path, along with a drone that can carry livestreaming equipment and sensors that thermally screen the environment. Operators can register the body heat of their target and the environment's AI animals resembling what pilots will experience on a search-and-rescue mission with drones.
Future improvements include a more lightweight final product, a better hiker AI, and better interfacing with QGroundControl.