Thinking about the limits of UAVs

Killing time in some less-than-inspiring conference (Hey! At least you get a free dinner!) I’m thinking about the limits of UAVs.

For example the new USN BAMS UAV will be a pretty complex thing with a new set of autonomous flight maneuvers.

And I was thinking that once you get the general flying part right, like a software library, you can apply it to any platform. Of course parameters will change, and the supersonic library is not really interesting for an infantry micro UAV. But the principle remains.

I think an important step would be to give the UAV a kind of 4D self-awareness during every part of the mission, including ground movement, and an awareness of its own need (like fuel – radioing for an AAR UAV) and vulnerability (evasive maneuvers when shot at, knowing that hitting the ocean will kill it). Multi-spectral spheric sensors, a rough “knowledge” of missions and environment, maybe some social behaviour patterns like software-ants.

But how far could automation/robotization go? BAMS will operate 24/7 with 3 vehicles. What if it could take-off/land autonomously, and you could do a MRO roboter unit that would refuel it and do routine maintenance jobs, being networked with the logistics base to call for spares (wich are delivered by autonomous cargo units, of course).

Not really Skynet-, more like HAL-capability, no real AI, but a very advanced expert system. Lost one? No problem – ask the bookkeeping software-agent at Congress for money, he says yes, hikes up the takes 0.1%, next year the factory will build one more XYZ, or send it over as replacement.

How far can such a scenario realisticly go? Homo Sapiens Sapiens units only as supervisors and for extraordinary events?

And I was also thinking of the influence a complete “fly.dll” could have on commercial operations. A single pilot cockpit, to make the pax feel good?