Will UCAVs be Smart Enough?

Posted 10 Nov 2001 at 23:31 UTC by steve

A new Business Week Online article gives a good summary of the X-45, one of several Unmanned Combat Air Vehicles (UCAV) under development. The article addresses the issue of reliable targetting. Are UCAVs really smart enough to distinguish a school bus from valid target like an enemy tank?

Missing the point..., posted 11 Nov 2001 at 18:54 UTC by josborn » (Journeyer)

I believe the article misses the point. It's trying to say that since the Army is now using "smart" bombers, those planes could improperly identify a target and destroy it. At least for now, the military isn't putting up drones and letting them work on their own, these are essentially big R/C airplanes with bombs on them and humans controlling them from the ground.

I highly doubt that the military will let these things go fully autonomous until the AI can decide on a target as well as a human controller.

R/C vs. Autonomous, posted 12 Nov 2001 at 04:16 UTC by steve » (Master)

That's true of UAVs like the Predator but the article is talking about stuff like the X-45 and Pegasus that are still under development. These UCAVs are designed to be flown by a ground control station and humans would have to authorize their attacks and arm the weapons systems, but the direct human control would mostly consist of selecting and executing scripts (sort of like on Star Trek when they order the ship to execute "attack pattern delta").

Most of the attacks involve the use of multiple UCAVs communicating with each other (and the group of UCAVs would usually have just one human operator). For example two UCAVs might be ordered to a particular area to destroy anti-aircraft weapons on the ground. One UCAV would identify targets and while the other UCAV flew in stealth mode firing weapons at the targets. Other strategies involve larger numbers of UCAVs working together.

After an attack plan is issued the vehicles are designed to carry it out even if all communications with the human operator is lost or jammed. They are supposed to be able to evade enemy fire, identify and destroy their targets, select new routes to get out of the target area if needed, and possible even engage manned enemy fighters (how much of that they can really do at this point is anybody's guess though).

So, it might be accurate to say they're much more autonomous than fighter planes with human pilots but not quite as autonomous as cruise missles. There was a Jane's article a while back that talked about some of this here.

they aren't stupid, posted 5 Dec 2001 at 16:33 UTC by Serac » (Journeyer)

I remember seeing that one of the planes, maybe the pegasus? Made quite a lengthy flight and handled take off and landing just fine without human control. I don't see how this technology could be a problem, with the level of AI programming even in games today i think that robot-military aircraft will most likely be very autonomous and better pilots than most human pilots in a few years.