Bug or Feature - A-10 Attack Logic

[Dusty Rhodes]

I was running tests on the A-10 and how far out it would attack using a target zone linked to a waypoint. I was a bit surprised by what I found. The convoy was running from east to west and was at a stand still, all T-80's.

 

When the Hog attacked from the south of the convoy (side) it would not attack unless the target zone waypoint was 4500 meters or less from the target.

 

When the hog attacked from the west of the convoy (Behind), it would not attack unless the target zone waypoint was 2500 meters or less from the target.

 

If it attacked from the east (In front) of the convoy, it would not attack unless the target zone waypoint was 3500 meters or less from the target.

 

If it attacked from the north of the convoy (side), it would not attack unless the target zone waypoint was 5500 meters or less from the target.

 

I thought what this could simulate is the difficulty of IDing a target from different angles with the side angle being the easiest way, but those distances, even from the side are way off of what a HOG driver could ID it's target. This pilot was rated as HIGH. I am going to do excellent next to see what the differences are. So my question is the attack logic whacked or is this done on purpose for ID purposes? Again, even if it is done for ID purposes, a good (HIGH) pilot could ID his target way beyond 5500 meters with his sensors and tactical knowledge of where he was on the battlefield.

 

Anyone from ED care to tackle this one? I'll be back after I test this with an EXCELLENT Pilot.

 

Below is the mission I am using. If you see where I have done something wrong, please point it out so I can be really embarassed and shut up :)

 

BETTER DATA WAS OBTAINED THRU FURTHER TESTING. READ THRU THE THREAD TO MY POSTS WITH THE CORRECT DATA AND MAKE OF IT WHAT YOU WILL.

Test Mission.miz

Edited February 4, 2009 by Dusty Rhodes

[Dusty Rhodes]

Well uping the rating to EXCELLENT did nothing. In some cases the ranges got shorter and some longer but definitely not in increments you could say was effected by the pilot rating. So now I am going to exchange my A-10 for an SU-25 and see what happens. This should be interesting.

[Dusty Rhodes]

Well I exchanged the Hog for the Frog and got a totally different story.

 

From the front of the column, the Frog ID"d and attacked from 11 thousand meters compared to the HOG's 3500 meters

 

From the rear of the column, the Frog ID'd and attacked from 10 Thousand meters compared to the HOG's 2500 meters.

 

From the right side of the column, the Frog ID'd and attacked from 11 thousand meters compared to the HOG's 4500 meters.

 

From the left side of the column, the Frog ID'd and attacked from a whopping 12 thousand meters, compared to the HOG's 5500 meters.

 

For both the HOG and Frog, the settings, speeds, altitudes, everything was identical except each had their own CAS Loadout with long range weapons aboard. It seems the HOG got shafted pretty bad in a head to head comparison. And the HOG's distances just don't seem right considering the HOG's equipment and the pilots tactical knowledge of where he is on the battlefield. I agree with the Frog settings because of the same reasons I disagree so vehemently with the HOG's settings. I hope this can be addressed and straightened out in the patch.

 

On to the helo's. Apache A and Longbow and Whiskey Cobra vs. Blackshark and Hind. Lets hope there attack distances are good to go.

[Dusty Rhodes]

Well I just finished testing the Helo's Attack and engagement distances and what I found was pretty interesting. The Helo's have a greater distance they ID and attack the targets from than the HOG and Frog. Here is what I found.

 

The Black Shark ID'd and attacked from 15 Kilometers and engaged from 7 Kilometers from the sides, front, and rear of the tank convoy which was stopped.

 

The MI-28 ID'd and attacked from 15 Kilometers and engaged from 5 Kilometers from the sides, front, and rear of the tank convoy which was stopped.

 

I did not test the Hind.

 

The Apache A Model ID'd and attacked from 15 Kilometers and engaged from 9 Kilometers from the sides, front, and rear of the tank convoy which was stopped.

 

The Longbow Apache mirrored the Apache A Model.

 

The Whiskey Cobra mirrored the Apache A and Longbow and when loaded out with TOW missiles, it engaged at 4 Kilometers.

 

So what we have here is helo's that have greater ID and Attack ranges and greater engagement ranges then the HOG and FROG. Again all tests were done using the same exact perameters for each helo and using the long range tank killer missiles of each country. It should also be noted that I used the M1A2 as the target for the Russian helos and it usually took 2 missiles to kill an M1A2. I used the T-72 and T-80 for the American targets and they blew up with just one hit except with the TOW missile which sometimes took 2 hits.

 

I hope ED takes a look at my test results and makes some adjustments, especially to the A-10. It has been short changed BIG TIME.

 

Now to test how many hits it takes for each Main Battle Tank in the sim from the Anti Tank Missiles. :)

[GGTharos]

There is no adjustments needed. You are not taking into account that units need TIME to also find targets with the range of their sensor. Because of the A-10 speed being much faster than a helo and only having eye balls to find targets, acquistion distances are much less as they should be unless the A-10 is told very specifically where to look.

 

Unlike LO, air units to not magically find enemies, they first have to spot them and the greater time they have the greater their chance of success.

 

The new AI aircraft detection logic is described in the GUI manual.

 

I hope this helps!

Edited February 3, 2009 by GGTharos

[Dusty Rhodes]

No adjustment needed? You think the ranges the HOG ID's at is OK compared to the SU-25 and the Helo's? I solidly beg to differ GG.

[Dusty Rhodes]

Well I did some testing on Main Battle Tanks and Anti Tank Missiles fired from Helo's and here is what I found:

 

From the front, back, or side, the Hellfire is one bad dude from the Whiskey Cobra, the A Model Apache, and the Longbow Apache. The only descrepincies here is that the frontal shot on the T-80 was 8 shots, 8 hits, and 6 kills and the frontal shot on the T-72 was 8 shots, 8 hits and 5 kills. Otherwise the Hellfire was perfect from the side shot.

 

The TOW, fired from the Whiskey was atrocious. Out of 16 missiles fired at 8 T-72's, then 8 T-80's, There were 11 hits, 5 misses, and 4 targets destroyed. I am a former TOW man myself with the ground unit and I remember how cranky that wire guided missile was. I think the hits and kills should be 50/50 but it is pretty close from what I can tell and my knowledge of the TOW system.

 

With the Black Shark the AI firing the VIKR scored 12 shots, 12 hits, and 6 kills on the M1A2 from the front and the side shots.

 

On the Leclerc, it was scored 12 shots, 12 hits, 5 kills from the side of the column and 12 shots, 12 hits and 6 kills from the front.

 

On the Leopard 2, it was scored 12 shots, 12 hits and 5 kills from the front of the column and from the side of the column.

 

On the Leopard1a3, it was 12 shot, 12 hits, and 10 kills from the front and 12 shots, 12 hits, and 9 kills from the side of the column.

 

So plan on shooting 2 missiles at the modern MBT and one for the older MBT's. I will report my findings on the Challenger 2 in a bug report I will be posting.

[GGTharos]

The max launch range normally attributed to the AGM-65 as far as I'm familiar with, and I'm sure a hog driver will pound me for being wrong if I am - is about 3nm. This isn't necessarily due to the seeker, and certainly not the propulsion of the weapon itself. It has more to do with the meat stick actuator.

 

If you're using the plain Su-25A for comparison, then there's probably something to deal with there, but if you're using the T, then no.

 

And as mentioned before; the sensor package of most helicopters as well as their speed affords them a better autonomous SA. And right now, the Ai are pretty autonomous AFAIK.

[Dusty Rhodes]

I was using the T in my testing. Now even if the T is spot on, and like I said, I had no problems with it's numbers, I have a problem with the numbers for the HOG. They should be much better then that though maybe not as good as the Frog-T.

[GGTharos]

I don't have any problems with it. They seem to be correct to me.

 

PS: The visual sighting ranges for ground units is taken from no-kidding pilot feedback (Tornado crews)

Edited February 3, 2009 by GGTharos

[71st_Mastiff]

Hey Dusty you weren't working for Microsoft FSX by any chance were ya?

That would explain why the game didn't sell well with your experiments a little faulty?

Don't take this in a wrong way I'm just really fascinated by your system and maybe your getting some anomalies with it? A Little jokey joke at ya too.

Thanks for putting this information out there and your time in doing it too, much appreciated.:thumbup:

[Feuerfalke]

Hey Dusty you weren't working for Microsoft FSX by any chance were ya?

That would explain why the game didn't sell well with your experiments a little faulty?

Don't take this in a wrong way I'm just really fascinated by your system and maybe your getting some anomalies with it? A Little jokey joke at ya too.

Thanks for putting this information out there and your time in doing it too, much appreciated.:thumbup:

 

So, what is your message? :huh:

 

 

Anyway, when the visibilty and ID-range taken from real pilots accounts is 3nm, why the differences between the Frog and the Hog?

 

Just to compare the numbers again:

Frontal attack: Frog T: 11km vs. A10: 3.5km

Rear attack: Frog T: 10km vs. A10: 2.5km

Side attack: Frog T: 11km vs. A10: 4.4km

 

And you say it's correctly modelled this way? :blink:

 

 

How comes the Frog T has 3-4 times the visibility and engagement range of the Hog?

[EvilBivol-1]

Because it has optics?

[mckee14]

interesting stuff :)

 

one may test what the ID-range/attack-range is, when the FROG and the HOG do not fly straigt in the direction of the targets, but instead fly paralell to them - so the AI has to detect the targets visually, and not by optics... that way, the SHOULD be equal

[bumfire]

Because it has optics?

 

But then, so does the A10, both 25T and a10 have tv screens, although the 25t has got the better zoom, but if both aircraft are flying high then with their screens both should be able to pick up and engage a target from more than 3km away, 3km isnt far from cannon range for the a10 nevrmind missile range, especially with the a10s IR mav's which will make targets easier to see and as ggt says, the mav's are quite capable of much further distances.

 

I think their is a huge discrepancy between the su25t and a10 in their attack ranges.

 

Although I havent tried to reproduce what dusty rhodes has, but with his numbers, if they are correct then I would tend to think something is off.

[Dusty Rhodes]

Just to be clear, I purposefully made it so the aircraft being tested had to turn to attack so I could tell when it was attacking. So it wasn't flying parallel to the target but at least a 45 degree angle to it prior to turning to attack. It was at such an angle that you couldn't use the optics to find the target. Since I am meeting skepticism, I will run the tests again today just to make sure the optics were not available to detect the target. If they are not, then the ranges should be the same, and I am confident that my first set of tests were set up so the optics couldn't be used. But I will run them again and post my findings.

Edited February 4, 2009 by Dusty Rhodes

[GGTharos]

But then, so does the A10, both 25T and a10 have tv screens, although the 25t has got the better zoom, but if both aircraft are flying high then with their screens both should be able to pick up and engage a target from more than 3km away, 3km isnt far from cannon range for the a10 nevrmind missile range, especially with the a10s IR mav's which will make targets easier to see and as ggt says, the mav's are quite capable of much further distances.

 

Look, this isn't the way it is. It is actually VERY hard to see a camouflaged (ie. appropriately painted) target from an aircraft, and it is difficult enough to see anon-moving vehicle on a road as well. If you haven't experienced it then I'm sorry to say that you don't have a basis of comparison for talking about it with those of us who have.

 

The maverick can be used as a 'poor man's FLIR', but it isn't. There are operational limitations for the missile, and despite the zoom capability it isn't an optical system for enabling detection and ID of target. It can simply 'kinda sorta' be used as one in IRL, and it's nothing like what you have in LOMAC. A-10 pilots use the Mk1 eyeball to acquire targets, and as such, at least AFAIK, even with Mav-D's the engagement distance ends up being not much more than 3nm, barring certain specific circumstances.

 

I think their is a huge discrepancy between the su25t and a10 in their attack ranges.

 

And there should be. The Su-25T can detect and identify targets from much farther out. It has hardware specifically designed for this.

 

Although I havent tried to reproduce what dusty rhodes has, but with his numbers, if they are correct then I would tend to think something is off.

 

No, it is not.

[Dusty Rhodes]

GG, I am going to run the tests again and make sure the optics are out of the picture for ID'ing the target. If I get the same results then YES, there is something wrong. I am confident I will get the same results.

[GGTharos]

Then I would say that what is wrong is potentially certain sensor slew capabilities, but in general I'd agree.

[mckee14]

edit: too late, and i guessed not bad i think :)

==============

 

my guess is, that it has something to do with these optics, so the SU25T has a way longer detection range. but then, i also guess, that the AI routines used to evaluate IF a target is detected or not, are kind of simplified (they are for sure, the question is how much?).

 

so maybe the SU25T has like a cone defined, starting on his nose, and is limited by mathematic factors like ["length_of_cone"=12000m; horizontal_angle=+/- 30degrees, vertical_angle=+0/-30degrees]. if a target is inside the cone and visible range is greater that distance of the AC to the target, and target is inside for longer than XY seconds (depending on the skill of the AI pilot), it gets "seen".

that way, it would be a good mix between realistic possibility of finding a target and not too much processor load...

 

but it can also generate some situations, in which a human player would have completely different results than a AI, or the results from plane to plane differs...

 

but only guessing here! imho, around 3000 meters seem to be realistic for detection by eye for single targets or small groups. bigger groups, especially when moving, should be detected from further awawy, depending on the situation.

[bumfire]

Look, this isn't the way it is. It is actually VERY hard to see a camouflaged (ie. appropriately painted) target from an aircraft, and it is difficult enough to see anon-moving vehicle on a road as well. If you haven't experienced it then I'm sorry to say that you don't have a basis of comparison for talking about it with those of us who have.

 

The maverick can be used as a 'poor man's FLIR', but it isn't. There are operational limitations for the missile, and despite the zoom capability it isn't an optical system for enabling detection and ID of target. It can simply 'kinda sorta' be used as one in IRL, and it's nothing like what you have in LOMAC. A-10 pilots use the Mk1 eyeball to acquire targets, and as such, at least AFAIK, even with Mav-D's the engagement distance ends up being not much more than 3nm, barring certain specific circumstances.

 

 

 

And there should be. The Su-25T can detect and identify targets from much farther out. It has hardware specifically designed for this.

 

 

 

No, it is not.

 

I know exactly how difficult it is to ID somethings when airborne since every 2 weeks I am infact airborne and get a nice oppertunity to look outside the windows on the way to and from work and see for myself what the world looks like when above whilst not being too high and travelling at a relatively low speed.

 

However that being said, the optical system on the 25t is just that, optical with a zoom function and the system on the a10 is also optical with the seeker of the mav being the camera and also has a limited zoom function, their is nothing magical about an optical system, ok granted the A10's isnt as good as a dedicated optical system that the 25t employs and is basically just a tv hooked upto what the seeker of the missile sees, but once the pilot ID's the target in the a10 he can then engage and then extend and comeback and on his second run engage at a further distance 3KM engage distance to me seems small in comparison to the 11 or so that dusty has seen. I am not saying its wrong, I am just saying its on the small side, maybe a release of mavs once the a10 has picked the target and is on his second run in at about 5km or so ? who knows exactly what the a10 pilots do ? I am sure they fire mav's at more than 3km.

 

Even if the engage distance was 3km or so for the first engagement then could they not increase the distance for the latter engagments with missiles only, I would think that would be more like it, ok I am not an a10 pilot and dont know how far the maximum range they will enage with mav's is but I would think if they had already eyeballed the target and knew where it was then they could increase the attack distance to more than 3km?

 

EDIT:

 

this is what I found

 

One major drawback of the A-model was the limited range at which the TV-seeker could lock on: although the missile has a range of up to 7nm (13km) under the worst possible circumstances (a low and slow aircraft), attenuation at optical wavelengths limits lock-on range to about 3nm (5.5km). Furthermore, the AGM-65A needs 4-8 seconds to lock on, which is an eternity on the modern battlefield.

Thats the a model which is TV guided, already we have a 5.5km lock on range and thats with the worst kind of maverick, go up to newer models and the lock on range increases thus so must the firing range once the target is found, and as you know yourself a pilot would rather engage the furthest possible distance from a live target once the target area and target have been correctly identified

 

There is also some confusion going on here, Dusty said that it was 3000 - 3500 metres engagement range, I agreed saying its kinda low, your reply was saying no its an ok range and that engagements are carried out at not much more than 3NM, 3NM is much more than 3KM, 3NM is roughly 5-5 n half KM which I said in my reply to you above that that would be a favourable distance to engage on a second run to the target. So we are kinda agreeing on the range, just that we are speaking KM and you posted in NM and have confused dustys original 3-3.5km as 3NM. Regardless, I am happy with 3NM + as that would equate to roughly 5.5km which I am happy with as an attack range. But that still leaves the question that if they can engage and normally do at 3NM give or take IRL then why are they engaging a full 2 - 2.5KM less in the sim ?

Edited February 4, 2009 by bumfire

[Dusty Rhodes]

Well GG, I was OVERCONFIDENT in my findings. My previous findings were a mix of in sensor range and not, so my results from the first tests were not correct in what I was looking for. The testing data I was looking for could not be found the way I did it previously faulty tests and for that I apologize.

 

With that in mind I ran a NO SENSOR POSSIBLE, Mk1 Eyeball only, from the side of the column, test and both the HOG and Frog had Mk1 capability inside 6 KM. They ID'd from inside 6 KM and attacked.

 

So then I did a series of test of FOR SURE SENSOR DETECTION, from the side of the column, and found the Frog to ID and attack from inside 16 KM and the Hog inside 6 K or the same as Mk1 Eyeball range, even with Mavericks loaded so the pilot had use of the TV.

 

I will do tests from the front for Sensor detection and from the rear sensor detection and post the results but right now I am eating a little crow for posting faulty data on the tests I ran before. I am now very confident, however, in what I found above and each person can make of them what they will.

 

BUMFIRE, in my tests, the Frog attacked from Max Firing range. The Hog got off what it could from the inside 6 KM is possible and then extends out and fires from Max range with great effectiveness. The AI attacks are logical and as they should be. It is the distances, which I ask that you read above, that was the data I was looking for. I still have to do rear and front tests, which I will get to in a while.

Edited February 4, 2009 by Dusty Rhodes

[bumfire]

so it looks as though they attack at greater than 3km range then ? within 6Km does sound alot better than 3000 - 3500 metres.

 

I like your methodolgy of testing, seems to be thorough, even if you did make a tiny miscalculation earlier, keep up the good testing DR :thumbup:

[Dusty Rhodes]

It wasn't a miscalculation but it was the angle that I had the aircraft flying to the attack waypoint. Sometimes it was too much for sensors and some times it was within. These tests I just did were no doubters.

[Acedy]

The maverick can be used as a 'poor man's FLIR', but it isn't. There are operational limitations for the missile, and despite the zoom capability it isn't an optical system for enabling detection and ID of target. It can simply 'kinda sorta' be used as one in IRL, and it's nothing like what you have in LOMAC. A-10 pilots use the Mk1 eyeball to acquire targets

 

W. L. Smallwood's book "Warthog" has two chapters about the "Night Hogs" (nice pun btw :D) flying night missions in the A-10A during Desert Storm and using the Mav for target detection and identification. Quite interesting read on what tactics they developed and the pros/cons of deploying the Maverick at night.