Lateral Missile Guidance Bug - AIM-54 and AIM-120B/C

[Whiskey11]

 Alrighty, so yesterday, I decided to do some testing with the Phoenix after the patch went live.

I noticed a long time ago that it seemed like the Phoenix TWS Trajectory doesn't make a lot of sense to me.  What I noticed is it would launch and go into a pure pursuit in TWS.  This, isn't a real big issue, except that it isn't actually calculating the pursuit angle correctly and actually makes an arc to the target rather than flying to the computed impact point.  This was contributing to the "self notching" behavior of the missile.  So here is my test parameters:

Map: Mariana's over the ocean pointing towards nothing but ocean

Launch Aircraft:
F-14B, ~33k Feet, Mach 1.21 AIM-54C, Course: Due North

Target Aircraft:
Su-33, ~36k feet, Mach 0.75, Course: ~210 degrees, Set to Not react (so no maneuvering, changing speeds, or deploying countermeasures, etc)

Distance at launch was 67.1nmi with the above parameters.

Here is a picture of the course the AIM-54 took  to guide in on the Su-33:

See the left hand arc in the trajectory?  Seems counter intuitive.  The Su-33 is not changing speed, not changing heading, nor changing altitude.  It's constant.  To me, the arc does not make logical sense.  Why would this missile not leave the aircraft and turn to the "perfect" intercept point to impact the target without making any other turns.  Again, the target is not maneuvering (no altitude or heading changes) or changing speed.  The impact point should be fixed in space such that the missile goes active without maneuvering laterally beyond the initial turn to the intercept point.  I should point out this problem ALSO exists in PD-STT at these same ranges although the steering que for launching in both TWS and PD-STT is different.  I'd also point out, that the computed impact point, FOR THIS SCENARIO, should not change at all from the initial one because the aircraft is not maneuvering or changing speed.  With that in mind, I don't think the TWS track refresh rate matters for this particular case.

Because I'm slightly crazy, I repeated this test with the AIM-120C and the F-15. The AIM-120C's lead angle on a ~45nmi shot is far closer to a straight line, but it still makes the turn.  

I also performed these tests with the F-16C (33k feet, Mach 1.56) with both the AIM-120B and the AIM-120C but the shorter launch ranges (~20nmi) showed a nearly straight line to impact, which makes sense given the time of flight of the missile.

I fully realize there is a "balance" between the missile making its initial turn and energy, but this seems like that computed impact point is very wrong for the AIM54/AIM120B/C and it is computing an impact point which is closer to the target than it should be.  The end result is a missile which is easier to notch because it's "behind" the pursuit curve. 

Hopefully Heatblur or ED can shed some light on the guidance issues seen here!  Thanks!  This is mostly a copy and paste of my post to Heatblur on the AIM-54 Feedback discussion, but in case this falls under ED's wheelhouse, I wanted you guys to be aware too!

F-14AIM54Test1.acmiF-15AIM120CTest1.acmiMissileGuidanceTest2.trkMissileGuidanceTestF14AIM54C.trkMissileGuidanceTestF15TWS120C.trkMissileGuidanceTestF15TWS120C2.trk

Edited June 17, 2022 by Whiskey11

[Маэстро]

That's not correct to directly compare AIM-120 and AIM-54 guidance, them use different API and different guidance laws.

There is no issue with AIM-120 guidance. It's ok if missile has almost straight trajectory but not exactly straight.

Regarding AIM-54 - such big arc is a result of guidance tuning of this particular missile. I will contact Heatblur.

[Hobel]

vor 8 Stunden schrieb Маэстро:

There is no issue with AIM-120 guidance.

I have a question about the steering.

Is there any way to give logic to the Aim120 and other missiles so that they don't head straight into the ground when the target aircraft is flying in that direction?

Even old systems like the SA-2 Sam have a countermeasure against such manoeuvres.

wouldn't it be quite realistic for modern rockets like the 120 to have such a logic?

thank you for your work

 

 

here is an excerpt with a time stamp

 

Edited July 13, 2022 by Hobel

[Whiskey11]

On 7/13/2022 at 5:55 AM, Маэстро said:

That's not correct to directly compare AIM-120 and AIM-54 guidance, them use different API and different guidance laws.

There is no issue with AIM-120 guidance. It's ok if missile has almost straight trajectory but not exactly straight.

Regarding AIM-54 - such big arc is a result of guidance tuning of this particular missile. I will contact Heatblur.

The only reason I used the AIM-120 was to see if it was a DCS issue vs a Heatblur AIM-54 issue.    Not trying to say they should be exactly the same.  The only way they would behave exactly the same is if they used the same guidance code in game, which they don't.

Heatblur responded to my post in their feedback thread for the patch stating they tweaked the amount of lead the missile pulls, but they are limited by the old missile API it is currently on.  Hopefully the AIM-54 can get on the new missile API soon and some of these issues can be addressed!  For a missile like the AIM-120 which is battery limited in terms of range, the guidance issue with having too little lead is a lot less of an issue, but when a missile like the AIM-54 can go 165nmi on a cooperative target (as it does currently), that angle builds up and leads to the self notching behavior.  The new API can't get here soon enough for the AIM-54  

Thanks Маэстро!!!

[Krippz]

On 7/13/2022 at 3:40 PM, Hobel said:

I have a question about the steering.

Is there any way to give logic to the Aim120 and other missiles so that they don't head straight into the ground when the target aircraft is flying in that direction?

Even old systems like the SA-2 Sam have a countermeasure against such manoeuvres.

wouldn't it be quite realistic for modern rockets like the 120 to have such a logic?

thank you for your work

 

 

here is an excerpt with a time stamp

 

 

I've been wondering about this myself. Right now one of the most effective methods at missile evasion is to just lead the missile into the ground. 

[okopanja]

On 7/13/2022 at 9:40 PM, Hobel said:

I have a question about the steering.

Is there any way to give logic to the Aim120 and other missiles so that they don't head straight into the ground when the target aircraft is flying in that direction?

Even old systems like the SA-2 Sam have a countermeasure against such manoeuvres.

wouldn't it be quite realistic for modern rockets like the 120 to have such a logic?

thank you for your work

Excelent question for just about any missile since at least 80s.

[Cmptohocah]

6 hours ago, Krippz said:

I've been wondering about this myself. Right now one of the most effective methods at missile evasion is to just lead the missile into the ground. 

How would the missile know it's heading into the ground?

[okopanja]

1 minute ago, Cmptohocah said:

How would the missile know it's heading into the ground?

Most of these have INS, coupled with initial position parameters, they can decide not to have interception point at several km of negative altitudes.

[KlarSnow]

If below a certain altitude point no lower than the target until a certain line of sight rate is reached.

So if in a low altitude environment, use pure pursuit in the vertical until the line of sight rate hits a point that requires switching back to lead.

Edited July 20, 2022 by KlarSnow

[Cmptohocah]

6 hours ago, okopanja said:

Most of these have INS, coupled with initial position parameters, they can decide not to have interception point at several km of negative altitudes.

That's all fine, but how would it know what "negative altitudes" are? To know which ones are "negative" as you said, it would have to know where the zero is. How would it know this?

 

4 hours ago, KlarSnow said:

If below a certain altitude point no lower than the target until a certain line of sight rate is reached.

So if in a low altitude environment, use pure pursuit in the vertical until the line of sight rate hits a point that requires switching back to lead.

 

If the launching aircraft is in "low altitude environment" or the missile?

[KlarSnow]

Could be either, the missile could have a "low altitude flag" that gets tripped at launch based on targets or shooters altitude. Or It could have some way of knowing its altitude roughly, if not accurately, even just barometrically. Or if its an active missile it could measure the altitude return once its active and have a very accurate altitude. Or if it has an INS/a 3d Coordinate and vector of the target it could switch the logic on itself based on the targets maneuvers and position.

Lots of options based on how specifically a missile works.

Could also just trigger this logic in the "terminal phase" instead of from launch, again based on any or none of the above factors.

[Cmptohocah]

13 hours ago, KlarSnow said:

Could be either, the missile could have a "low altitude flag" that gets tripped at launch based on targets or shooters altitude. Or It could have some way of knowing its altitude roughly, if not accurately, even just barometrically. Or if its an active missile it could measure the altitude return once its active and have a very accurate altitude. Or if it has an INS/a 3d Coordinate and vector of the target it could switch the logic on itself based on the targets maneuvers and position.

Lots of options based on how specifically a missile works.

Could also just trigger this logic in the "terminal phase" instead of from launch, again based on any or none of the above factors.

Altitude and height above the ground are two different things. Altitude is just a pressure measurement and has nothing to do with the terrain unless you are over the sea in which case "ground zero" is well defined, as the sea surface. You can test this out by flying on Caucausus map and setting visibility to near zero. Then fly by checking your baro altimeter only: chances are you will hit terrain at some point due to mountains.

Once it's active it has its antena pointing at the target so unless it has a radar-altimeter there is no way to tell its height above the ground.

INS tells you your position relative to some reference point but this also does not provide you with your height above the ground. It would need to get the ground level info from the launching aircraft or from its own RA, or have a map with topological features which it would read from, or a combination of the above.

[KlarSnow]

Yes I am aware, that does not change any of the validity of it. The vast majority of terrain in the world is within a few thousand feet of sea level, and if you are ducking into a valley there isn't much a guidance algorithm can do to avoid a ridge. However that's not what we are talking about. We are talking about a target at low altitude, in clear line of sight of the missile diving at the ground to push the intercept under the terrain. That happens most often relatively close to sea level. So having something like when below 5000 ft AGL barometric, activate this logic, would solve 90% of these issues. Missiles have to have some SA as to the density altitude they are flying into because it can change how they actuate their fins, so it is a part of their logic to change gains based on shoot up, vs shoot down.

No it would not solve this issue 100% of the time, but you need some pretty advanced stuff to solve this problem 100% of the time. And again nothing is going to help the missile avoid a ridgeline its target dives behind, however over relatively flat even terrain this would prevent the majority of the exploits that occur in DCS. (how much of each map has terrain that isnt mountainous, and is above my example altitude of 5,000 feet MSL)

The only one that gets remotely close is the Nevada map in some parts. And even then, most of it is below 4000 ft MSL.

 

As to requiring a radar altimeter, if the missile has an active radar of its own it will recieve a strong altitude line through its sidelobes, any MPRF or HPRF radar will have that, and especially as it gets closer (within a mile or two) it should be fairly easy to track the range to it. This would be a direct measurement of AGL.

Edited July 21, 2022 by KlarSnow

[KlarSnow]

Example from SAMSIM's documentation on the SA-2

Target height less than 5Km (15-16,000 feet), at least this variant of the SA-2 uses exactly the method of guidance I am talking about to avoid the ground. That's a pretty big buffer and would also solve the issue immediately.

[Frederf]

F-16 has terrain database which in theory could provide a detailed terrain value but I doubt it. Currently selected steerpoint elevation is plenty good or even just assuming system altitude = height so don't go for negative height. For half a million dollars per missile one might want to put some brains in it. Launch altitude would be a valuable data to load before launch for all kinds of reasons.

[Cmptohocah]

10 hours ago, KlarSnow said:

Example from SAMSIM's documentation on the SA-2

Target height less than 5Km (15-16,000 feet), at least this variant of the SA-2 uses exactly the method of guidance I am talking about to avoid the ground. That's a pretty big buffer and would also solve the issue immediately.

Is the missile guiding by itself or is it being guided by the SA-2 system?

If latter is the case, than it should not be very difficult to tell the AGL since the entire system sits at ground zero.

10 hours ago, KlarSnow said:

Yes I am aware, that does not change any of the validity of it. The vast majority of terrain in the world is within a few thousand feet of sea level, and if you are ducking into a valley there isn't much a guidance algorithm can do to avoid a ridge. However that's not what we are talking about. We are talking about a target at low altitude, in clear line of sight of the missile diving at the ground to push the intercept under the terrain. That happens most often relatively close to sea level. So having something like when below 5000 ft AGL barometric, activate this logic, would solve 90% of these issues. Missiles have to have some SA as to the density altitude they are flying into because it can change how they actuate their fins, so it is a part of their logic to change gains based on shoot up, vs shoot down.

No it would not solve this issue 100% of the time, but you need some pretty advanced stuff to solve this problem 100% of the time. And again nothing is going to help the missile avoid a ridgeline its target dives behind, however over relatively flat even terrain this would prevent the majority of the exploits that occur in DCS. (how much of each map has terrain that isnt mountainous, and is above my example altitude of 5,000 feet MSL)

The only one that gets remotely close is the Nevada map in some parts. And even then, most of it is below 4000 ft MSL.

 

As to requiring a radar altimeter, if the missile has an active radar of its own it will recieve a strong altitude line through its sidelobes, any MPRF or HPRF radar will have that, and especially as it gets closer (within a mile or two) it should be fairly easy to track the range to it. This would be a direct measurement of AGL.

 

Ok, so maybe I took the wrong approach here - in theory it's possible, sure.

Do you know any A2A missile that has this sort of capability? 

[KlarSnow]

The SA-2 radar has no idea what the AGL under the target or intercept point is. That is the entire point of using the LOS to the target as the boundary. if I keep my intercept no lower than where I see the target now, unless terrain gets between me and the target then the missile cannot hit the ground.

and in this case it really doesn’t matter if the missile is doing the guiding. In the SA-2’s case the guidance commands are sent from the radar to the missile, the guideline missile doesn’t guide itself, but this simple line of sight based adjustment still works if the missile is doing it all on its own. 

All you need to do is pre launch tell the missile you are going into a low altitude environment, (whatever you define that as) or the missile tells itself when it crosses a specific altitude to swap. 
I do not know if any specific air to air missiles that use this specific technique, but I do know that the AIM-7E and F had altitude bands that were passed pre-launch based on the targets altitude in order to set control gains.

Whether or not the AIM-7E/F had this type of technique to help with low altitude targets I don’t know, but the pieces are all there and it’s not a difficult problem to solve especially with computers.

Edited July 21, 2022 by KlarSnow

[GGTharos]

On 7/21/2022 at 5:24 PM, Cmptohocah said:

Ok, so maybe I took the wrong approach here - in theory it's possible, sure.

Do you know any A2A missile that has this sort of capability? 

Solutions to this problem have existed since the 50's, and the easiest one is as far as I know KlarSnow pointed out ... fly at an elevation slightly above the target (or same as, depending on application) then go PN or whatever other form of terminal guidance is desired on that axis when the LOS rate on that axis reaches a specific threshold.   This is used to solve a number of low-altitude-target issues, not just the 'my target is diving into the ground' thing.  So, you don't really need any fancy measurements for the most rudimentary approach - the only data you need is the missile's vector, and the target's LOS rate from the seeker.   You could even do this with a heat seeker if you wanted to for some reason.

[AeriaGloria]

On 6/9/2022 at 11:40 AM, Whiskey11 said:

 Alrighty, so yesterday, I decided to do some testing with the Phoenix after the patch went live.

I noticed a long time ago that it seemed like the Phoenix TWS Trajectory doesn't make a lot of sense to me.  What I noticed is it would launch and go into a pure pursuit in TWS.  This, isn't a real big issue, except that it isn't actually calculating the pursuit angle correctly and actually makes an arc to the target rather than flying to the computed impact point.  This was contributing to the "self notching" behavior of the missile.  So here is my test parameters:

Map: Mariana's over the ocean pointing towards nothing but ocean

Launch Aircraft:
F-14B, ~33k Feet, Mach 1.21 AIM-54C, Course: Due North

Target Aircraft:
Su-33, ~36k feet, Mach 0.75, Course: ~210 degrees, Set to Not react (so no maneuvering, changing speeds, or deploying countermeasures, etc)

Distance at launch was 67.1nmi with the above parameters.

Here is a picture of the course the AIM-54 took  to guide in on the Su-33:

See the left hand arc in the trajectory?  Seems counter intuitive.  The Su-33 is not changing speed, not changing heading, nor changing altitude.  It's constant.  To me, the arc does not make logical sense.  Why would this missile not leave the aircraft and turn to the "perfect" intercept point to impact the target without making any other turns.  Again, the target is not maneuvering (no altitude or heading changes) or changing speed.  The impact point should be fixed in space such that the missile goes active without maneuvering laterally beyond the initial turn to the intercept point.  I should point out this problem ALSO exists in PD-STT at these same ranges although the steering que for launching in both TWS and PD-STT is different.  I'd also point out, that the computed impact point, FOR THIS SCENARIO, should not change at all from the initial one because the aircraft is not maneuvering or changing speed.  With that in mind, I don't think the TWS track refresh rate matters for this particular case.

Because I'm slightly crazy, I repeated this test with the AIM-120C and the F-15. The AIM-120C's lead angle on a ~45nmi shot is far closer to a straight line, but it still makes the turn.  

I also performed these tests with the F-16C (33k feet, Mach 1.56) with both the AIM-120B and the AIM-120C but the shorter launch ranges (~20nmi) showed a nearly straight line to impact, which makes sense given the time of flight of the missile.

I fully realize there is a "balance" between the missile making its initial turn and energy, but this seems like that computed impact point is very wrong for the AIM54/AIM120B/C and it is computing an impact point which is closer to the target than it should be.  The end result is a missile which is easier to notch because it's "behind" the pursuit curve. 

Hopefully Heatblur or ED can shed some light on the guidance issues seen here!  Thanks!  This is mostly a copy and paste of my post to Heatblur on the AIM-54 Feedback discussion, but in case this falls under ED's wheelhouse, I wanted you guys to be aware too!

F-14AIM54Test1.acmi 117.08 kB · 8 downloads F-15AIM120CTest1.acmi 136.88 kB · 8 downloads MissileGuidanceTest2.trk 1.21 MB · 5 downloads MissileGuidanceTestF14AIM54C.trk 678.31 kB · 6 downloads MissileGuidanceTestF15TWS120C.trk 880.74 kB · 5 downloads MissileGuidanceTestF15TWS120C2.trk 555.22 kB · 4 downloads

 

Isn’t this just advanced proportional navigation? Is the Phoenix not suppossed to have APN but collision course nav? APN is one of the best forms of navigation for maintaining missile kinetic energy 

[Whiskey11]

3 hours ago, AeriaGloria said:

Isn’t this just advanced proportional navigation? Is the Phoenix not suppossed to have APN but collision course nav? APN is one of the best forms of navigation for maintaining missile kinetic energy 

There was actually a bug in which the AIM54 was not using PN at all but a continuous lead angle.  This was fixed in Open Beta a patch or two ago and guidance has improved tremendously.