Forums » Air Combat & Civil Aviation » DCS World » Flaming Cliffs / LOMAC » FC3 Confirmed. Active Topics You are not logged in. [Log In] [Register User]
Page 9 of 19 < 1 2 ... 7 8 9 10 11 ... 18 19 >
Topic Options
Rate This Topic
Hop to:
#3469649 - 12/07/11 12:25 PM Re: FC3 Confirmed. [Re: Nate]
Flogger23m Offline
Senior Member

Registered: 12/06/02
Posts: 2733
Loc: California
The AI can use some work. There were some improvements with FC3, but I hope more is done in this area in FC3. Many times the AI fails to do what I would consider simple things.


Top
#3469655 - 12/07/11 12:38 PM Re: FC3 Confirmed. [Re: Nate]
Nate Offline
Member

Registered: 05/24/01
Posts: 1217
Loc: Dublin, Ireland
Flogger if you could detail the problems you experience I may be able to get them sorted, but no promises. The more detail the better along with short tracks would be cool.

I spent quite a bit of time testing the Wingman AI in BS2 - which i believe worked out well - I'll try to do the same for the FC3 AI, but any help I get in highlighting issues would be much appreciated.

Nate

Top

#3470104 - 12/08/11 06:29 AM Re: FC3 Confirmed. [Re: Nate]
zzzspace Offline
Member

Registered: 09/26/11
Posts: 131
Loc: Australia
Originally Posted By: Nate
Originally Posted By: zzzspace
Quote:
Please post a detailed account of your problems (maybe even a track or 2) in here and I'll see what I can do.

http://forums.eagle.ru/showthread.php?goto=newpost&t=81721

Nate



I'll do that Nate, a track or two may be best, with a description of what I'm trying to acheive with that flight.


Cool thanks, note if the tracks are done in Warthog or BS2, they will most accurately reflect the most current state of the AI. Also try to keep the tracks as short as possible, so I can easily highlight the issues to the team.

Nate


I was referring to the AI in FC2.1, and only in regard to it's pending upgrade to FC3.0, so I'll do the tracks in FC2.1.

I made some tracks last night using SEAD Tornado and SEAD F/A-18C plus SEAD F-16C over mountains) that fully displays these problems, but these are not brief tracks, as you have to let these flights fly detailed low-level paths with about 20+ waypoints, down valleys. All the waypoints were locked to 75m (without exception) and all aircraft were set to excellent/expert pilot settings, in order to see fully what it is doing wrong.

I'll try to describe it clearly for you Nate;

1) The AI tends to attempt to jump to the altitude of the next waypoint, as soon at it passes the last one. i.e. it does not do ANY actual low level flying near 75m AGL -- whatsoever! So if transiting over a mountain spine, the next waypoint may be thousands of meters higher than the last waypoint. So instead of the AI actually flying a low level profile @ ~75m AGL, it just immediately climbs (often using burners and wasting fuel for nothing .., and keeps doing it, even if you reduce the speed requirement between the waypoints!) so it just cruises to the next waypoint at that waypoint's altitude plus 75m (unless there is an intersecting ridgeline or high-point in between), thus flying thousands of feet above ground level!

[ i.e. A FAIL OF LOW LEVEL RADAR AVOIDANCE TACTICS]

2) The second problem is separate, but occurs on top of the first problem. Notice I said "unless there is an intersecting ridgeline or high-point in between". i.e. when the AI does fly to intersect a higher peak or ridge in between waypoints, regardless of the aspect of the slope it intersects (be it 5 degrees or 75 degrees) the AI will always immediately closely approach to the ground, then it will pull-up very hard, at about 70 degrees from horizontal, at full afterburning power, then it climbs upwards literally 3000 to 5000 feet AGL! Then it will nose over at the top and enter a shallow or steep dive, towards the next waypoint's elevation, plus 75m again!

[ i.e. A FAIL OF LOW LEVEL RADAR AVOIDANCE TACTICS]

So there are two altitude tracking problems that are overprinting here Nate. But let me continue before I say what I think causes this;

3) The AI afterburning climbs that are occurring during all this nutso behaviour is quite gratuitous. The AI seems to be blind to its approaching high terrain until it has almost hit it. Then it suddenly wakes up (it 'panics') and instantly goes to full throttle, and max pitch climb. It doesn't attempt to 'see' ahead at all, to manage/regulate both the thrust and climb or decent angle, pre-emptively, with respect to the actual slope and altitude of valleys or ridges intersecting your planned flight-path's altitude profile, to achieve a smooth and efficient flight path result -- i.e. unseen by almost all enemy radars, and even AWACs or fighers.

--

Indeed, after watching Lockon 1.02, FC1.1, and FC2.1 I've come to the startling conclusion that Lockon doesn't actually have ANY genuine low-level or else terrain-following flight algorithm or capability--whatsoever!


It seems this was (somehow) overlooked in development and put on the backburner, and this is why it simply attempts to jump to the next waypoint's altitude, instead of even attempting to tracking the terrane profile along its flight path to the next waypoint.

What we have instead seems to be merely a development place-holder, that somehow became permanent.

PLUS, we have a crude terrane-avoidance knee-jerk 'reaction' occurring, and simply as a developmental work-around for the fact that there is no actual terrane-following or low-level flight algorithm.

I know that in either the first or second LOMAC v1.0 patches, a terrain avoidance 'fix' was applied, and I think what we're seeing here is this same 'fix' ... and it actually fixes nothing!

It only prevented collisions with the steepest terrane, but it also totally destroyed any chances for exploiting the radar shadow for low level tactics ... it was only done because the aircraft kept plowing into the steep terrane within LOMAC v1.0, so this fix simply made an ALREADY existing developmental placeholder 'reaction', of the AI discovering terrain close, at short notice, to pull-up even harder, and to use more power for longer, and thus to climb even higher over ridges, before nosing over at the top, to then adjust to the elevation of the next waypoint.

This is all applied to cope with the fact that there's no actual low-level terrane-following or low level AI flying capability in Lock ON FC2.1 -- at all.

The other thing is the way the AI behaves (it just dicks-around) near to the target, rather than simply striking hard and fast, then getting out ASAP. Keep in mind, these air craft were all set to Excellent expert level.

That's what's going on Nate, and I believe these problems are affecting ALL the aircraft, not just the SEAD flights, because all of these have no low-level terrain following flight model.

If this is rectified Nate it will simply totally transform Lockon FC -- very much for the better.

I believe tens of thousands of people would flock back to FC, and its sister products, if you got this low-level flight operations working properly so that what you planned tactically, actually was executable, in practice (I salivate at the planning, mission and campaign prospects!).


EDIT: spelling and layout.


Edited by zzzspace (12/08/11 06:42 AM)
_________________________
||| Romanes eunt domus ||| zzzspace V2.0 REAL SOUND for DCS World - and all Modules |||

Top
#3470126 - 12/08/11 08:06 AM Re: FC3 Confirmed. [Re: zzzspace]
Nate Offline
Member

Registered: 05/24/01
Posts: 1217
Loc: Dublin, Ireland
Originally Posted By: zzzspace


I believe tens of thousands of people would flock back to FC, and its sister products, if you got this low-level flight operations working properly so that what you planned tactically, actually was executable, in practice (I salivate at the planning, mission and campaign prospects!).


Not so sure about thousands flocking to LOMAC over this smile Thanks for he detailed description and I'll certainly take a look at those tracks (upload them to mediafire or similar). The reason for mentioning Warthog or BS2 is that is generally the AI version that will be in FC3 and is generally much more flexible in how it handles waypoints etc.

Again I'd like to point out that I'm not promising a fix at all, it may prove to be too difficult to get working within the project time constraints, but it certainly isn't being ignored.

Nate

Top
#3470691 - 12/09/11 04:00 AM Re: FC3 Confirmed. [Re: Nate]
zzzspace Offline
Member

Registered: 09/26/11
Posts: 131
Loc: Australia
I'm very glad you guys are going to have a close look at this Nate. I've all the other DCS titles installed, but not had time to play them.

Been busy building an FC2.1 soundpack lately. The sound-engine is big step forward.s I've identified multiple new ways to exploit its potential. What I ended up with is stunningly close to the real thing. In fact, close your eyes and it is almost indistinguishable from the real thing, most of the time. No longer sounds like a computer game trying to sound like the real thing, it actually sounds like it might be the real thing. It's pretty damn cool and will only get better.

I'm working all weekend and (hopefully) a couple of days off after this, so I'll get some track files and a test mission link up for you by Monday or Tuesday.

And sorry Nate, I really do think thousands will flock back to lockon if it actually does work at a tactical planning and mission execution level, as planned.

You will see! yep


Edited by zzzspace (12/09/11 04:26 AM)
_________________________
||| Romanes eunt domus ||| zzzspace V2.0 REAL SOUND for DCS World - and all Modules |||

Top
#3470720 - 12/09/11 07:01 AM Re: FC3 Confirmed. [Re: Nate]
Nate Offline
Member

Registered: 05/24/01
Posts: 1217
Loc: Dublin, Ireland
Note DCS has a selection to set waypoints "above ground level" rather than just the "Mean Sea Level" only available to FC2. This will make waypoints much easier to place and use compared to FC2. Also the latest DCS editor has copy and paste (CTRL-C, CTRL-V), so flights with hundreds of waypoints can be duplicated and repositioned easily.

In general I'll request again that you use the newer versions of DCS for tracks - the ME is better, more detailed, and the AI is the most current version. These are the versions of the ME and AI that FC3 will be based upon.

Nate

Top
#3471848 - 12/11/11 08:59 AM Re: FC3 Confirmed. [Re: Nate]
zzzspace Offline
Member

Registered: 09/26/11
Posts: 131
Loc: Australia
Nate, I've done this now with DCS A-10C v1.1.0

I've not patched to a higher version, and given what I saw and recorded I see no reason to do so, as it was a major Ai dud from the low-level tactical planning and execution perspective. The attempted execution of planning by the DCS was a debacle. The results recorded were actually worse than FC2.1, Which surprised and disappointed because I thought (and hoped) I might need to eat my words. But there's no danger of that. It simply made my points more clearly.

Lockon FCx and DCSx has no genuine low-level fight modellimg. This emporer has no cloths on. What is there is a disastrous series of hopelessly failing placeholders, that very inadequately cope with the lack of any genuine low-level flight capacity.

This is ED's feet-of-clay.

It's the part they should have sorted out, from the very beginning but didn't. It now undermines the evolution they've achieved so far. I don't know why it was ever permitted to go on this long, or to fail so spectacularly at what the mission editor implicitly claims it was able to do, but can't.

It doesn't work even at a basic and rather simple level.

The closer you get to the ground, the greater the planning execution failures become, and the more numerous and spectacularly stupid the result becomes.

I will say in fairness that the AI did try harder in DCS to remain at a lower level. Someone did try to make this work better. But this in fact made it's failure worse. Much of it is due to the VERY --- V E R Y -- crude built-in terrain avoidance knee-jerk's as over-reaction with full burners plus max-pitch climb almost vertically (or more!) for 1,500 to 3,000 feet AGL -- seen by every radar in kingdum-cum!

I did like the DCS open-formation, but NOT for transiting! This should only occur at the IP waypoint. As the tracks will show and open-formation transit is a major disaster for low-level flights, down valleys etc, and this made the DSC AI's overall failure more glaring. In fact, it caused several crashes into terrain, in short order.

Also, one DCS SEAD flight, of four, COMPLETELY IGNORED ALL OF THE PLANNED WAYPOINTS!!

Instead it simply flew from the take-off point, in a straight line, to attack the nearest enemy radar emitter! This creates an exposed approach providing enormous early-warning of the attack.

FAIL

Also, the treatment of waypoints as ASL or AGL, in DCS make no real difference, and is not an issue to consider (and it was AGL in LOMAC and FC anyway prior to DCS). The ASL option really only helps for things like looping aircraft to keep them stable in transitions over water towards high terrain, or to keep aircraft above effective MANPAD AGL and aspect. ASL doesn't matter or help for low-level flight plan execution, as then the AGL primarily matters.

Basically DCS AI added nothing useful in terms of needed solutions to past plan execution problems. But is did demonstrate some new and serious flaws within the AI's addressing of targets and it's fundamentally suicidal (and very half-a$$ed) engagement habits (I won't call these tactics).

I'll write-up some notes for each track (four of them) and post links, probably tomorrow. Basically, DCS's AI was astonishingly abysmal at low-level combat, and even hopeless at recon. It failed in every significant respect to execute the most prosaic of low-level tactical flight, plus attack, recon, target addressing, egress and simple RTB requirements.

The inefficiency and shear unnatural-ness and stupidity of flight manoeuvres and the wastage of fuel was absolutely ridiculous. These were mostly due to over-reactions, and also under-reactions to the terrain. It's nothing short of disgraceful.

What I expected to find, after about 7-years of development, is a mature evolved combat sim, in all core respects. What I found was the reverse.

I can't find words sufficiently scathing for the mess that is the DCS AI low-level tactical flight.

I will say ED has done a spectacular and successful job of mid to high alt air-combat simulation--I can't fault it for that. And battlefield engagement is well done -- not great -- but OK.

But this sim is deeply and painfully flawed at a paradigm and also structural level. To think that a helicopter and A-10C low-level attack sim has no functional low-level fight modelling ... is simply beyond my capacity for suitably scathing remarks.

The AI requires a major re-think. And also the manner in which the mission planner deals with flight performance, in terms of turn-radius's and the climb rates at a given planned speed and load, so that the mission planner itself can check the terrain profile along the track, and the flight performance of the aircraft, to determine in advance in planning, if a planned waypoint placement is going to be 'flyable', and achievable, at the speed, altitude AGL (and also ASL as pressure matters), and for the load-out configuration planned for the assigned role and estimated fuel burn rate.

These things HAVE to be a part of the mission planner, or how else is the aircraft going to be able to achieve what is planned? Or else to let the planner know in advance that it won't be feasible to do that route in that way?

This is what any real attack mission planner has to take into account - and so should the DCS FCx mission planner.

Without that, and an integrated low-level flight model capacity, that can meet and mirror what the mission planner has calculated and determined, in advance, to be achievable, then Lockon DCS FC3.0 will turn out to be yet another low-level tactical strike and multi-role engagement failure.

Of that I'm sure.

These are not things I want to say at all, but they certainly are well-past-due to be said, and in blunt terms.

As it stands, anything ED generates from here, with this rendering of the DCS AI, can only sabotaged FC3.0, via the AI's absurd-esque failures to fly literally A N Y low-level mission plan -- as planned.

The DCS mission planner and the DCS AI are fundamentally incompatible.

This can not and must not go on.
_________________________
||| Romanes eunt domus ||| zzzspace V2.0 REAL SOUND for DCS World - and all Modules |||

Top
#3472970 - 12/12/11 10:43 PM Re: FC3 Confirmed. [Re: Nate]
zzzspace Offline
Member

Registered: 09/26/11
Posts: 131
Loc: Australia
These DCS A-10C v1.1.0 low-level flight experiments were kept deliberately simple and prosaic. Nothing complicated was attempted and basically the AI takes-off then flies over a coastal plain, over a few small hills, up a steep valley and over a dam wall and lake, then over a few ridge lines connecting classic deep glacial valleys, at a suitably low-level to avoid early detection and warnings by medium range GBAD radar. All waypoints in the flight are thus set to 65m AGL (~200 feet AGL). This is a very low level of flight, but strike pilots are typically trained extensively to do this consistently. I of course realise the Ai in this program, nor any human or aircraft can fixate rigidly on maintaining that sort of height, in practice (though pending death would help). But what I do expect and what should occur is that the Ai will try to smoothly and efficiently attempt to approximate a 65m AGL within the smoother valley floors, and also to not exceed 65m AGL by much as it noses over any high ridgeline, and to be back well below it again within a few seconds.

I wanted to see if the Ai would again begin to fly over the dam lake at a steady 65m AGL, and it did, but as soon as it reached the end of the lake and met terrain again any semblance of low-level flight became a debacle, leading to crashes into terrain, and the ludicrous 'heavy-metal yo-yo' behaviour as the AI repeatedly over and under reacts to the changing contours of rising and falling terrain.

The flight plan follows the valley WNW for about 150 km and the flight emerges from terrain masking, back toward the edge of the coastal plain near the SAM site, thus unmasking from cover at waypoint 16, and either attacking, or else RECON of the SAM site, then a rapid dash back into steep terrain. I tried this exact same rout and SAM placements for SEAD, CAS and RECON modes with Western and Russian types, to see the effect of weapons on the Ai.

All the aircraft in these tracks were set to EXCELLENT AI setting, and the opposing SAMs are set to AVERAGE AI level. The tests were done with a Tornado (1 & 2), Su25T (3) and Su34 (4).


TRACK NOTES:



TRACK 1 http://www.mediafire.com/download.php?325zlbv6mj8m0cd
The first track shows a 4-ship Tornado EXCELLENT SEAD flight armed with ALARM and Sidewinders, that takes off then completely ignores the mission flight plan's waypoints and instead flew almost directly on a track to the nearest (Buk) emitter and attacked it from about 2,000 feet up (despite being set to 65m), where upon all 4 aircraft were systematically destroyed. Some of the SAM units were destroyed. There were no tactical 'pop-ups' of pincer manoeuvres used by the Ai, to find it's targets and fire, then dive back towards radar degrading clutter, and terrain-masking. The Ai RWR should sense the emitter type and thus fly an appropriate arc to avoid a direct pass over or too close approach to the SAM (Buk, with TORs and Tunguskas nearby). Instead the SEAD aircraft simply flew straight in at about 1,800 feet AGL toward and over multiple active SAM systems, in full and continuous radar view. Naturally all aircraft got wasted. Thus tactical target 'addressing' was a complete FAIL also, and thus egress and RTB observations were not applicable. FAIL



TRACK 2 http://www.mediafire.com/download.php?815j26xrrxhp2oa
This attempt worked only slightly better -- i.e. extremely badly. It is exactly the same scenario as Track 1, except this time I removed the ALARMS and other weapons and set the Tornado to EXCELLENT RECON role. So a 4-ship RECON flight takes off and does attempt to follow the mission waypoint plan this time. The usual absurd yo-yo routine as aircraft leap vertically up to 5,000 feet AGL soon develops. But it's worse than what you see in LOMAC and FC1 and 2, because the DCS aircraft fly an open formation in transit, so only the lead aircraft flies down the actual path plotted through the valley, so it the only one able to meet the time speed and alt requirements. The rest are all over the shop and the flight becomes very disorderly, disconnected and uncoordinated. Thus it can not reach the waypoints as planned and tactical surprise and coordinated time-compressed attacks are also completely out of the question with such a chaotic nonsense of yo-yo-ing aircraft. They are highly visible to any EW or AEW or even fighter radars, much of the time, thus completely defeating the point of planning a low-level undetected approach. And keeping ultra low most of the time is the only way you're going to minimise detection and break tracking by AEW and vectored fighters.

The manic yo-yo-ing finally stopped only when one crashed and the others finally ran low on fuel while still about 120 kms short of their closest approach to the recon 'target' area. Yeah, that's right, they got about 50 kms along the flight path before they ran out of fuel, caused by a combination of the aircraft not having the flight performance envelope necessary to reach the next turnpoint, due to the ridiculous afterburning climbs to avoid a ridge line that is soon >2,000 meters below!

Pure FAIL insanity - DCS - digital COMBAT simulation

The mission planner provides no feedback of the intended flight path's consequences regarding topographical profile changes directly along the planned flight path transect, and it does not calculate the turn radius arc and pitch radius arc sizes required to pass smoothly at about 65m (instead of >2,000m) above the ridgelines. The mission planning system itself needs to be forward-looking, even as the plan is being manually created, in order to anticipate and calculate to and inform the mission builder something like;

"This aircraft type at that height and temp, with that load-out, at that speed, in that flight formation, can not make the turn radius requirement to the next waypoint, or else it can not pitch sufficiently to nose-over the next ridgeline. Please manually move the waypoint further away until it can, or else allow auto adjustment to speed, load-out or waypoint location (in that order) so the aircraft can make that turn.

Do you want to auto adjust now? [Yes] [No], etc.

The mission planner should simultaneously recalculate if the aircraft is going to have sufficient fuel to complete the planned waypoint route for the attack and RTB. If no, the fuel should be increased, or waypoints edited, or else speed reduced, until it can achieve RTB at the planned base, with a sufficient fuel reserve buffer for emergency evasion needs.

I personally think an emergency evasion from actual direct attack by fighters or SAMs, or from mechanical failure and damage are the only valid reasons for ANY aircraft to autonomously disregard flight plan waypoint details. In which case a low level flight SHOULD STILL KEEP FLYING AT LOW LEVEL, and the aircraft should RTB as a soft 'mission kill'. You should be informed of its failure to achieve its planned task.



TRACK 3 http://www.mediafire.com/download.php?9fyu1bah2n6p1b6
With this track I took exactly the same scenario as the first two and simply replaced the Tornado with the Su25T, and swapped the Russian SAMs for NATO medium and short range Hawk, Avenger and Chaparral units in unchanged locations. I gave the Su25T EXCELLENT CAS role, with no weapons. The track developed in much the same way as Track 2, except the combination of the Su25's lower performance and the open formation within a narrow valley led to it being more prone to collide with high terrain. But the three surviving aircraft eventually simply gave up on terrain-following yo-yo-ing altogether the moment they flew high enough (thousands of feet above a ridgekine) to be detected by the Hawk SAM radar once in range of it. Thus once an aircraft is detected by the Hawk SAM radar it then flies level at about 2,000 feet AGL!

The completely wrong tactical response!

But the aircraft that are disordered and have fallen behind due to the open formation keep on yo-yo-ing until they also are individually detected, then they too fly level at about 2000 feet AGL in view of the SAM system. Now common sense would tell a pilot this is a particularly dumb thing to be doing right in front of a Hawk battery, especially when there's a deep valley right below you.

But the DCS AI is indeed depressed, at it's general inability to fly low-level, so is ready to end it all.

Thus the remaining aircraft fly toward the SAM in lower mid-level flight, totally ignoring their 65m AGL mission planning waypoint alt level, plus they're now strung-out over a 6 to 7 km distance, due to the yo-yo nonsense. So they independently trundle towards their dooms, with the airbase's SAM defences whereupon they are one after the other blown out of the sky without further ado. Good riddance. But if there were AEW and fighters around, they would not have made it even that far anyway.

Turkey shoot = SYSTEMATIC Ai FAILURE



TRACK 4 http://www.mediafire.com/download.php?82y8dw9uy29rl8j
Exactly the same as Track 3, except I replaced the Su25s with Su34 in EXCELLENT RUNWAY ATTACK role, with iron bombs to see if they could fly any more sanely, but they were even worse than the others, with three of them crashing into terrain, and the fourth did so many afterburning yo-yo's that it simply ran out of fuel, aborted to the nearest runway. This is an aircraft with a loaded combat range of several thousand kilometres!!! But it could not make a 200km low-level attack because of the staggering inefficiency of the way the aircraft attempt to fly at low-levels.

Another sad and pathetic EPIC FAIL of low-level attack flight tactics. I considered making another track with the A-10C, but what's the point, the whole thing is the mother of all cluster effs.



--

What I'm most appalled by is to see a supposedly dedicated low-level flight attack combat sim, in this ludicrous state of terrain following and target addressing dysfunction. This is actually supposed to be a Hi-Fi simuilation of low-level ground-attack! WTF?!

I won't dwell on that, but I suggest to the developers of FC3.0 that if a flight is set to fly below 200m AGL in the planner that formation should automatically change from open formation to line-astern formation to prevent the sort of transitory discombobulation and carnage that results from open formation transits at low level. And if set to go above 200m AGL by all means use open formation.

I would also add that the ground attack aircraft still actually need a pre-attack IP waypoint where the AI switches from a line a-stern transit NAV formation, to an open attack formation (STILL AT LOW LEVEL!!!) and that attacks should remain from 65m AGL so that a multi-axis pincer of low-level arcing/flanking manoeuvres (say 45 degrees either side of the target, from the IP), and flying brief tactical pop-ups and turns-toward the target manoeuvres to quickly find targets and fire weapons, then to pop-down into the weeds again and turn parallel away again of the beam, to minimise exposure, plus make the SAMs work much harder for their lock and kill.

The aircraft should make one or at the most two concerted attacks, in quick succession to minimise total exposure time, each with multiple SEAD weapons released, then immediately fly directly toward the next planned waypoint, at very low level to egress and recombine the flight in a line-astern formation at the next waypoint for a low-level RTB (an AEW, fighter and SAM evading withdrawal).

Aircraft should on no account swan-about @ 2,000 feet AGL, within the vacinity of defended targets. I really don't know why I nor anyone else has to explain these things or has to take the time to explicitly spell them out. If ED does not know the state of this sim's AI behaviour they are simply O U T O F T O U C H

If they know about it then why is it still like this? The Ai developers simply aren't reality-checking their output, or else are not permitted to develop them further by the Producer.


Tactics and the manner of target addressing matters greatly. The classic-case of low-level SEAD tactics being successfully used is the IDF's F-4 operations over the Bekkar Valley against Syrian SAM sites, in1982. The Syrian SAMs were almost completely wiped-out while the IDF lost zero SEAD attack aircraft. This sort of tactic is what I expect to be able to routinely plan for and execute within Lock On FC2.1 or DCS. Since LOMAC v1.0 to the present day, this series has been totally incapable of doing anything even remotely like this. Within DCS it actually got more dysfunctional.

An excerpt from: “FLIGHT INTERNATIONAL” - interview with Israeli Airforce regarding the 1982 "air war" over the Bekkar Valley eastern Lebanon (3832, Nov 16th1982)

F.I.: What types of electronic warfare were used?

IAF: You don't expect us to tell you that, do you? Electronic warfare was used all the place, all the time, by both sides.

F.I.: Was the Hawkeye effective?

IAF: It was used all the time. It can close the gaps in ground radar and lets us see the targets as they take off, never mind where they are within 200 miles.

F.I.: How did you out the SAM sites?

IAF: It is difficult to tell you, for publication. Since we've not finished the war against the SAMs, all the ideas, tactics and weapon we have for them. Knocking them out was much simpler than some of the press reports imply. If I told you it was done with conventional iron bombs you wouldn't believe me. But that is how we did it - with bombs, COMING IN VERY LOW AND DESTROYING THEM ONE BY ONE. It is true we used many deception techniques, decoys and electronic warfare to help us get in, but we destroyed them with bombs.

F.I.: How did the Syrians counter your attacks?

IAF: They have the most complex and dense SAM system in the world: SAM-6, SAM-3, SAM-2. They are three or four Gun Dishes on every battles ( Gun Dish is the fire control radar used with both SAM-9 and the ZSU-23-4 self propelled quad-23mm AAA system) plus 23mm, 37mm, and 57mm AAA. They knew we were going to attack. They welcomed it, they were ready. The only thing is they put too much faith in their SAMs. Maybe they were misled by their initial success in the 1973 war. In the first few days then, we had to fight the Egyptian armour crossing thw Suez Canal and were vulnerable to the SAMs. But afterwards we destroyed 43 SAM batteries on the Egyptian side.

IF THE SYRIANS HAD STUDIED THAT WAR WELL, THEY WOULD HAVE SEEN THAT WE KNOCKED THEM OUT WITH THE SAME METHOD AS WE DID TODAY. The Syrians have invested 75% of their defense budget on ground-to-air defense. They've got three times as many SAM batteries as in 1973. They have 80,000 of their regular soldiers on SAMs, only the best - educated ones. They have 1,000 tanks without crews as a result. I don't know … they were so sure that we'd knock our heads on the SAMs, that they would work as the Russians said they would. Maybe they were led to believe that mobility would solve their problems.

They forgot that with mobility you forsake the protection of being dug in under cement cover, like they used to be. So when they are mobile they're above ground. Once above ground, A CLUSTER BOMB WILL KNOCK OUT A WHOLE BATTERY, SO THE ONLY PROBLEM IS TO FIND THE BATTERY.

F.I.: What about the use of drones against the SAMs?

IAF: We did use drones, but not specifically against SAMs. Drone use was exaggerated by Press reports. They're in the evaluation stage. It is a new weapon. We use drones throughout the war , that we used them on the days we knocked out the SAMs too, but not specifically for than reason, mainly to locate ground forces to attack them. But we used RPVs in all kinds of ways. We used them as decoys and for reconnaissance. And we used mini RPVs (IAI Scout). But the used of drones was exaggerated by the Press . I believe that we really don't yet know what it means to operate such a weapon.

F.I.: Are SAMs too highly rated?

IAF: I've mentioned the 1973 war. I don't know if you know, but in that war the Syrians and Egyptians together launched some 2600 SAMs. They hit only 39 of our aircraft. The rest of the 102 aircraft we lost were hit by AAA. By the way, they also shot down 45 of their own aircraft. The capabilities of their SAM system were exaggerated. They were led to believe that they were defended, with a clear sky, under the SAM umbrella. That's why they invested so much in the system. But in this war they lost 100 aircraft and all their SAMs in Lebanon. I believe the Syrians in a dilemma today.

F.I.: Then Russians are also in an air defense dilemma?

IAF: Yes.

F.I.: Is it equipment or an operator problem? If Israelis had manned the SAM batteries, would the result have been different?

IAF: I don't think so. They made several mistakes but they did he best they could with the equipment they have. I believe that deficiencies of SAMs are such that, for us, it will never be a primary weapon.

F.I.: Why not?

IAF: It's a passive system. You invest money and manpower in a system which sits and waits for an air force to fly over it. An aircraft can be used for other purposes. I quoted the launching 2600 missiles in the 73 war. We launched 40-50 Hawks and killed 22 MiGs.

[CAPS my emphasis]

--


The key point above is the IDF were able in 1982 to routinely use terrain masking, "COMING IN VERY LOW AND DESTROYING THEM ONE BY ONE", through the valley terrain to take out whole SAM sites and the only complication was, "TO FIND THE BATTERY", and to blind it, rendering it ineffective.

As it turned out (and this article does not mention it) the IDF first used a very low-level approach and anti-radiation missiles, from F-4s, to take out EW and search radars, then Kfirs used pop-up "toss bombing", (possibly laser-guided and supported by special forces) from terrain-masked areas, to suppress lingering defences, that were flushed out via drones and decoys, then the IDF flew attack aircraft right in at very low levels and dropped cluster weapons on what remained of SAM site infrastructures.

--
See also: Operation Mole Cricket 19 - June 9, 1982

The Mastiff RPVs went in first to cause the Syrian SAMs to turn on their radars by convincing the Syrians that many attack aircraft were overhead. Once the Mastiffs were tracked by Syrian radar, the tracking signals were relayed to another Scout outside of the missiles' range. The Scout then relayed the signal to E2C Hawkeye aircraft orbiting off the coast. The data gathered was analyzed by the E2Cs and Boeing 707 ECM aircraft.[25] Once the SAM crews fired missiles at the drones, the F-15s and F-16s provided air cover while F-4 Phantoms attacked the SAM batteries, destroying them with AGM-78 and AGM-45 anti-radiation missiles.[26][28] The rapid flight time of the missiles minimized the F-4s' exposure to the SAMs.[1] The Syrians reportedly fired 57 SA-6s, to no effect.[29]

According to Ivry, many of the SAM batteries were on the Syrian side of the border.[2] Said Eitan, "From the operational point of view I can say that we used the mini-RPVs, long before the war, to identify and locate all the Syrian missile batteries. We then used superior electronic devices which enabled us to "blind" or neutralize the missile sites' ground-to-air radar. We rendered them ineffective to take reliable fixes on our aircraft aloft. But in advance of direct aerial attacks, we used long-range artillery".[30] The Syrians responded by launching about 100 fighter aircraft to stop the attacks.[2]
http://en.wikipedia.org/wiki/Operation_Mole_Cricket_19
--

As you may have realised, all the elements to acheive thios sort of lossless attack on sams and hardened targets is present in FC already. EXCEPT:

1) Ai has no ability to approach attack or egress in cover at low-level
2) Ai addresses and approaches the target in a completely suicidal way
3) Ai fails to egress immediately after a concerted but brief attack
4) Ai flies in extremely silly and inefficient ways, wasting fuel for no reason, and failing to RTB along the planned evasion route

Very low-level flight, in terrain masking mode, using SEAD weapons and ECM and drones is how you destroy such SAM systems. Lock On FC2.1 and DCS FC3.0 should be able to both plan and execute such very low level attacks, and suffer zero or nearly zero losses when planned and executed correctly.

All of this detailed terrain, detailed radar shadow modelling with clutter/noise degradation, and detailed modelling of SAM performances, but no capacity to use classic and prosaic low-level flight and attack tactics to address the needed target types, means you get a tactically senseless turkey-shoot every time. Not all attack aircraft fly day-or-night in any weather at low level, but those that were designed to definitely should.

At present there's no incentive to use the mission editor to plan strike attack packages as it just turns into a silly arcade-style shoot-'em-up that's not even remotely representative of actual air-to-ground and ground-to-air combat.

But if this all worked as it should then it would then be up to the individual planner/ mission commander and lead pilot to develop and use their skill and planning smarts to work out how to use what's available to best effect to destroy all of the enemy targets, and also survive.

In which case, instead of being given a finished mission plan (as FC does now in missions and campaigns) you could instead be presented with a defended primary target's location and (imperfect) recon intel, and asked to develop a tactical plan to kill the targets with minimal or no loses. This I would find vastly more engaging and enjoyable and provide an incentive to learn and get it right, because then its all down to how well you plan and fly and execute with coordinated timing.

Add a modern multi-role flyable supersonic low-level interdiction precision strike aircraft (not a mere battlefield CAS aircraft) and you've got combat sim nirvana, where any possible scenarios can be developed and planned out, and spectacular and tactically meaningful campaigns generated.

For me, if FC is not moving in that concerted direction, it's going to remain a low-level tactics joke.

This AI mess has gone far too long and I see no evidence that Eagle Dynamics has made any serious attempts to address these glaring and self-condemning deficiencies that I (and others) pointed out in detail to Matt and Karl, about six years ago. I can even provide links to the posts where this was previously discussed. They have known about these problems all along and have done nothing at all about it. It's clear to me now that ED is not serious, and has not been for a long time, about providing a realistic combat simulation. It flouts the terms DCS (digital COMBAT simulation) but it is just playing with 'techie' geekish details of avionics, sensors, comms and weapon's systems. The combat sim part itself is very much a secondary or tertiary consideration of the DCS producers, but it's actually the primary purpose for the customers.

In the absence of actual detailed developments aimed directly at addressing these crucial low-level air-combat simulation failures, as a fundamental core realism issue for ED, rather than just giving us more cheap vague verbal assurances of potential but nebulously inexact evolutionary progressions to come one day ... in a galaxy far, far away ... that's all this is being "considered" or else "seriously looked into", ... but alas, ... how sadly, ... it may not make it into this particular patch, or into this upcoming rendition, called FC3.0, due to production schedules, priorities, commercial constraints, and resource limitations ... oh blah, blah, f@cking-blah ... !!!

I've heard it all before.

I can even provide links to where such reassurances were given in the past when these things were discussed with ED. So no offence Nate, but I've heard these fob-off lines before, they amount to zip. Apparently that's the normal response that issues from ED to any real criticism of the state of things, when Matt in actual fact has no intention of ever doing anything about it. Firstly, you get a professed shared-concern, and collegial sincerity for getting these issues looked into and dealt with, but hey, we don't know precisely when this will occur, but do have faith in us dear customer, its already in the developmental pipeline, or soon will be.

Except that it isn't, and it never has been.

ED just says they're looking into it but they don't. Been there and done that Nate. So I'm rather cynical and out of patience for what Matt and ED spokespersons say, especially when they're always hedging, and failing to produce any improvement to the underlaying Ai catastrophe, but keep making 'new' sexed-up versions of the same underlaying dysfunctional tactical krud, with a few more ancillary (nice to have but non essential) baubles. The mission planner keeps getting expanded and reviewed, but the AI is never able to do what the mission planner can plan.

I can imagine how demoralised the mission editor programmers must feel to see such a system fatally hobbled by an AI that's incompatible with executing planned missions--and it's NEVER fixed!

Given all this, and the time that has elapsed I see no compelling reason to buy another Eagle Dynamics product as I see no evidence ED is in any way serious about creating a realistic tactical air combat simulation, with the core tactical aspects of low-level strike and interdiction combat.

I distinctly remember the first time I read the feature-list for LOMAC v1.0 and thinking - WTF?! No multi-role strike aircraft are on this feature list? What?! But we were all assured that plans were in the works toward that end, and all in good time little grass-hopper, al in good time ... well that time is well and truly up Matt.

$hit - or get off the potty!

In the low-level flight multi-ship package attack regime you only produced a shallow spectacle of generic air-battle 'cheese'. Actions not words Matt, because nothing you can say at this point is going to convince me you mean a damn word of it, or will deliver. ED is apparently pre-occupied with launching more and more high-brow commercial plans for this same underlaying broken-a$$ Ai, resulting in the usual tactical drivel with a new specular gloss. While everyone is still apparently cheaply quelled by the logical 'necessity' to go down this path, rather than make what exists, actually work properly. It's ridiculous, and it has gone on for far too long.

No more slack-cuts or giving you the benefit of the doubt. You've comprehensively blown it and will have to work very hard to get any of my money in future.

And to the DCS FC3.0 Beta testers;
I just want to convey this; If you're not going to hold ED's feet to the fire to create a tactical combat sim that actually functions and fights in realistic tactical ways, what exactly are you doing?

Does your definition of combat 'realism' mean you're prepared to agonise over avionics modes and such, whilst just accepting this absurd situation of a protracted thoroughly unrealistic tactical low-level combat that is almost completely dysfunctional?

I would hope not but why is it like this then -- still? I don't want any clever excuses. You're just being fobbed-off repeatedly by Ed's commercial production process and decision-making, that prioritises the development of ancillary gadgetry, that's all ultimately irrelevant if the underlaying tactical sim itself develops the air combat fight in a totally unrealistic and fundamentally porked way. You just passively accepting this absurd situation. Why are you content to let this sim remain in a complete mess like this? What's in it for you? So inside track news on the next version? So sense of collusion and attachment to these growing failures?

Look at those tracks and try telling me I'm wrong if you dare. What are you letting Ed off the hook so easily, giving them a free ride? Competition of ideas leads to excellence, not fawning sycophantic behaviours. It's necessary for creative and passionate people to push improvement for it to happen. If you don't push for it nothing will change. And almost nothing has changed in 6 to 7 years in terms of the parlous state of the low-level tactical combat, as described above1. Face it, Ed has gone stale, lost the creative spark, and so have you. You are simply not militating to add pressure for action to get these things addressed. You are being fobbed off by superficial small concessions and verbal gestures and flattery by ED.

And that's in part why it's STILL in a god-awful mess. You may even mostly fly on-line, and think the Ai is not so important to you, but even those SAM systems and ground forces are all Ai, and many of your opponent's attack flights are also Ai, and they can't execute low level plans, so you are not experiencing a low-level combat sim, but just the shabby appearance of it. So even in on-line play these systems can not and do not operate in realistic tactical combat modes, or roles.

Don't you want ED to fix this tactical mess? Why work on a 'new' version that doesn't fix any of this? I put this sim aside when FC1.1 was released because I was so disappointed by how it was performing. ED had basically ignored all of the issues discussed above, during FC1.1 development, even though these had been discussed at length openly prior to FC1.1 being even announced. Still, I regularly checked back on development to see if or when things moved forwards and matured, and was hopeful from reading glowing one-eyed reviews about FC2.1 that things had finally actually improved a lot.

But they haven't.

It's the exact same dysfunctional tactical Ai mess as LOMAC 1.0 and FC1.1, with almost ZERO significant improvements. Yes, there are minor but welcome tweaks here and there, but it's still completely porked as a tactical combat simulation. The lack of applied pressure is, in-part, why ED is still selling multiple versions of the same underlaying lemon to people who don't even comprehend how absurd it is in a low-level flight regimes and tactical employment regards, because they have not even known any sim that actually worked properly in those areas, or else are now habituated to the unrealistic stupidity that is FC and DCS.

Have a look at those tracks, or better still, make some yourself, then act on these observations. Be direct, don't just play-nice with ED, to avoid ruffling feathers. These guys need to be ruffled up as they aren't going to get the message loud and clear any other way. They have no passion for it any more. You know as well as I do that this comes down to ED's executive producer being out to lunch on these issues. He's clearly out of touch, or else blind to it all. He apparently thinks that the fact that a large chunk of the sim doesn't even work, is no biggie, no obstacle of selling more of this fault-ridden rubbish to more unsuspecting punters.

I won't respect or defer to that.

You may just see yourself as a beta-tester, of whatever ED condescends to provide to you for testing and reporting back observations, but you're collectively in a strong position to vigorously discuss and push ED to pull it's finger out, stop making the same lame old excuses and deliver a fixed Ai, and a low-level terrane following flight mode, capable of executing what's been planned so that stealthy strike attacks are both do-able, and survivable.

After all these years, do you still think ED (or Matt in particular) will EVER just get all this sorted out, one day, by themselves via making cumulative sequential "the right" commercial decisions and playing it safe?

What bollocks.

It just isn't going to happen via that route--and so it hasn't. Matt will of course always come up with the next excuse as to why ED can't fix it right now, but will continue to offer the just over the horizon possibility of it eventually occurring ... in some great celestial commercial whet-dream time scenario ... in the fullness of time.

It's a fob-off.

ED's producers have by implication demonstrated that either they totally fail to grasp what's fundamentally wrong, or else, and I this is very much more likely, that Matt Wagner simply does not give a toss whether this 'sim' does not work, in any realistic tactical manner. His multi-year lack of prioritising the necessary resources and action to make it work, so that a mission plan is executed as planned (WYSIWYG missions), is self-evident and standing proof of this.

As are the tracks provided. I'm tired of endless excuses for this total failure and all the idiotic justifications for why it needs-be so, and I'm too peeved by the almost complete lack of forward progress after so many years, to be patient any more. Nor to buy into any more of this fetid promotional titillating swill coming from ED about how FC3.0 will make a significant difference to Eagle Dynamics' underlaying feet-of-clay, and the notion that it is a combat sim, par-excellence.

New blood ED.
_________________________
||| Romanes eunt domus ||| zzzspace V2.0 REAL SOUND for DCS World - and all Modules |||

Top
#3473165 - 12/13/11 08:58 AM Re: FC3 Confirmed. [Re: Nate]
zzzspace Offline
Member

Registered: 09/26/11
Posts: 131
Loc: Australia
Eureka Nate!

A very promising breakthrough / discovery.

I found an aircraft that does fly low-level in terrain-following mode, as planned, and it does it very well.

The B1-B

TRACK 5 http://www.mediafire.com/download.php?j7ar0c8a570foae

I set the B1-B to PINPOINT STRIKE with EXCELLENT AI (although I suspect the role and Ai level does not alter the low-level flight behaviour, but need to check).

It all works for the B1-B, flies just like I want, as planned, and as I expected it could, and should.

I discovered this with FC2.1, and then tried it also in DCS and it works equally well in both sims.

I suspect it may even work for all of the heavy bombers in each but have to check that.

It may even work in LOMAC 1.02, but I've yet to try that.

This demonstrates that there is fully-functional low-level flight code in FC2.1, already, but the other aircraft are not using it.

Or rather, they are not using it properly.

I then studied the F/A-18C, using the exact same mission, and waypoints, and determined that these also are in fact using a terrane following code, but they're massively over reacting to the terrain.

i.e. the terrain sensing parameter and reaction trigger is simply set way to sensitively for fighters and attack aircraft.

The next track #6 shows that they are in fact trying to do the same thing as the B1-B, but they are just constantly over and under reacting to the terrain map. They react much too late, and when they finally do react, they massively over-react with pitch and power, then are much slower to cut power and nose over, like the B1-B does.

That's what's going wrong with it all Nate.

If this is fixed it will solve most of the problems, then it is a matter of refining the way the AI addresses the targets, with attack and egress tactics that are realistic, and it may all work.

You need to get your guy to isolate the code for the B1-B's terrain avoidance then macro apply it to the other aircraft, and there's a good chance this will be the necessary solution to this low-level flight disaster.

F/A-18C

TRACK 6 http://www.mediafire.com/download.php?c7yq5cp3qumwh6d
_________________________
||| Romanes eunt domus ||| zzzspace V2.0 REAL SOUND for DCS World - and all Modules |||

Top
#3476138 - 12/17/11 08:31 PM Re: FC3 Confirmed. [Re: Nate]
zzzspace Offline
Member

Registered: 09/26/11
Posts: 131
Loc: Australia
Only ED informal feedback on these issues thus far:

http://forums.eagle.ru/showpost.php?p=1354374&postcount=136

http://forums.eagle.ru/showpost.php?p=1354730&postcount=144

--

Naturally, imposition of trade sanctions has become necessary.
_________________________
||| Romanes eunt domus ||| zzzspace V2.0 REAL SOUND for DCS World - and all Modules |||

Top
Page 9 of 19 < 1 2 ... 7 8 9 10 11 ... 18 19 >
Topic Options
Rate This Topic
Hop to:

Moderator:  EinsteinEP, Panther, RacerGT 
 

Forum Use Agreement | Privacy Statement
Copyright 1997-2013, SimHQ Inc. All Rights Reserved.