"phil hunt" wrote in message
. ..
On Thu, 18 Dec 2003 22:25:54 +0000, Paul J. Adam
wrote:
In message , phil hunt
writes
On Thu, 18 Dec 2003 05:26:01 GMT, Kevin Brooks
wrote:
That is way beyond even our capabilities. You are talking autonomous
combat
systems.

Yes. The progrsamming for this isn't particularly hard, once you've
written software that can identify a vehicle (or other target) in a
picture.

Falling off a cliff isn't a problem once you've learned how to fly like
Superman.

Trouble is, that prerequisite is harder than you might expect.

Getting a machine to tell a T-72 from a M1A1 from a Leclerc is hard
enough in good conditions

You don't have to. You have to be able to tell whether it's a
vehicle or not, and if it is, is it in an area likely to be occupied
by own forces.

That 'area' changes hourly. And may not be known until the weapon gets over
the target area.



: doing so in the presence of camouflage,
obscurants and when the crew have run out of internal stowage (so have
hung lots of external gear) and maybe stored some spare track plates on
the glacis front ('cause they need the spare plates and they might as
well be extra armour) gets _really_ tricky. Do you err on the side of
"tank-like vehicle, kill!" or "if you're not sure don't attack"?

I'd tend to err towards the former. note that it's a lot easy to
spot a moving vehicle than a stationary one.

Do you waste a missile on a dark green Chevy Suburban, or a tank? Do you
have enough missiles for *every* vehicle in front of you?


Would it not be embarrasing to have a successful armoured raid broken up
by your own missiles?

Indeed. Maybe some form of IFF?

Even the US/NATO gets that wrong sometimes.

Pete