Re: [OT] Performance of Simulators

I just gave a paper at SIMTECT 2004, the Australian Conference
on simulator technology.

While waiting to give my paper, I had time to hit the M1 Tank
training simulator and RBS-70 simulator.

After 15 minutes on the M1, I was getting a 1st round Pk of 50%
with sabot, but a lot less with HEAT. Average time from target
appearing to destruction was 15 secs. That was with laser operational,
but auto Fire Control off. It would have been a lot easier if there
had been less cover - with no clear LOS, only an antenna to aim at
through the shrubbery, you can't get a good range with a laser.

Ranges were anything from 700 metres to 4 km. Close in, the high
rate-of-change-of-azimuth for crossing fast moving (60 km/h+)
T-80 targets was tricky, especially as they zigged.

With the RBS-70 - well, I got two hits out of 20 attempts.
The major problem was the coarse tracking. Once I'd managed to
get the thing trained onto the close vicinity of the target,
steering it to hit was comparatively easy. But the simulator
had the Z-controls for the fine and coarse tracking reversed -
pull back the thumbsticks to go down. Bleah!

Although the graphics were quite realistic (as opposed to a
single moving dot on a black screen), it was very similar to
the SeaCat simulator I'd used back in 1965.

It occurs to me that a simple edge-filter tracker could be made
for the RBS-70 quite easily. Get the pipper close to the target,
tell it 'go fetch', and it will automatically slew to centre the
target on the scope. Given a good range and doppler from the 
laser, and the calculations involved to get a minimum-energy
intercept (or even a straight constant bearing one) would be
trivial. It would mean that gunner skill wouldn't be quite so
important, especially with total novices.
Not good vs targets in NOE, nor in bad weather. But in normal
conditions, it would raise the Pk dramatically.