Bringing study-level air traffic control simulation to web with Unity
All the world's study-level simulators run on-site on offline simulators, typically on computers marked specifically for that use. There has never been a need to make them accessible over the internet — training is typically done on site, next to an instructor.
For hiring, this is a matter of scalability: requiring applicants to come on site for an assessment reduces overall capacity, which with it comes the problem of selecting who to invite. So, the ability for all applicants to do an assessment at home on any device, even the cheapest of Chromebooks, is necessary to eliminate at least one point of selection failure.
With this comes our first challenge. Existing "consumer" simulators (read: video games) do not present an accurate representation of an air traffic control environment even close to study-level. Neither do these simulators have scoring features more sophisticated than detecting whether two planes have collided. Autonomous simulators fit for instruction must have rule-breaking detection algorithms.
To solve these problems, we have developed the Separation Integrity Engine (SIE) alongside our basic terminal simulator. The SIE can detect losses in separation across a variety of scenarios that extend beyond the basic 3 miles horizontally and 1000 feet vertically basic separation minimum.
This first generation of SIE — which before now have not been implemented on any air traffic control simulator — will detect 99% of all possible errors in the terminal environment.
Why Unity?
Unity's WebGL export target allows us to compile the full simulation to a format that runs natively in any modern browser. This means candidates need no downloads, no plugins, and no special hardware beyond a browser and a headset — the same barrier to entry as any modern web application.
The simulation runs at full fidelity in-browser, with radar position updates, flight data block management, flight strip systems, and all voice input processed in real time through the cloud inference layer.