2021 A Falcon BMS Year to remember!
-
I hope that before any new features are introduced, all current bugs will be solved, namely the following:
- All impact timers of GBU’s are messed up, they either speed up or slow down and even give feedback if a weapon is guiding properly.
- IAM’s weapons time of flight is way too long and contains very weird slowing down phenomena. Just compare for example a release of a CBU-87 with that of a CBU-103 at the same time and look at the flight profile and especially the velocity of the weapon. For DEAD operations this is very crucial.
- Released weapons having a large horizontal shift upon release in cross wind.
- Release point of AI aircraft for GBU’s is way off compared to human pilots and bomb TOF is also very different.
- AGM-88 estimated impact time and countdown timers for PET shots are way off and even give feedback if the weapon is guiding on something. And often a HARM will not guide onto an active emitter if it has not been detected by ownship before.
- AI fuel consumption is way off compared to human, especially low- and high level.
- AI should not use afterburner when rejoining in flight.
- AIM-120 behaviour is unrealistic. RCS setting in cockpit determines exactly at what distance from the target the missile goes pitbull and that distance does not depend on actual RCS at all. The RCS selection in the pit should only provide an indication to the pilot about when the computer in the jet thinks the missile is going pitbull depending on the expected size of the target.
-
@bowser
1 > - AI fuel consumption is way off compared to human, especially low- and high level.
2 > - AI should not use afterburner when rejoining in flight.1 - Please elaborate.
How did you evaluated this?
Have you considered AI (robot) behavior and leader abilities to be a good leader considering the nature it his wingmen?2 - Please provide factual examples.
Also, consider that is could not be possible to make a difference between rejoin and other flight phases … so … if we prevent IA from using afterburner, it may be all also prevent them from using is in combat …
See the problem ? -
I was told all the mentioned bugs were fixed already for 4.36, only the RCS thing needs to be checked/verified.
-
@bowser said in 2021 A Falcon BMS Year to remember!:
- AI fuel consumption is way off compared to human, especially low- and high level.
AI fuel consuption is off indeed (i wouldnt say WAY off ) , but actually the AI burns less fuel than the player , didyou think it was the opposite ?
-
And the explanation is more simple than what i thought :
the Player and AI are calculating EXACTLY the same fuel flow, independantly of OFM or AFM
But at the end of the calculation there are those lines :
// For simplified model, burn less fuel
if (IsSet(Simplified))
{
currentEngine->fuelFlowSS *= 0.75F;
}“Simplified” in the code does not mean a simplified FM, it means a simplifed handling mode , for instance when flying formation or going to waypoint this mode is activated.
When in combat mode, this is no more “simplified”
So…i think this line has been added long time ago to compensate from the fact the AI was not as smart as player to manage Throttle when rejoining and so on…
i dont think we should touch it actually
-
@bowser said in 2021 A Falcon BMS Year to remember!:
- AIM-120 behaviour is unrealistic. RCS setting in cockpit determines exactly at what distance from the target the missile goes pitbull and that distance does not depend on actual RCS at all. The RCS selection in the pit should only provide an indication to the pilot about when the computer in the jet thinks the missile is going pitbull depending on the expected size of the target.
sorry but we do not understand what you mean ,
According to our information the way we modeled the RCS setting in SMS is realistic
the RCS option in the SMS determines at what range the Seeker will go HPRF / MPRF, independantly of course of the real RCS of the opponent
it also determines the size of the signals that the missiles radar will be looking for
we double check and the implementation in BMS is correctly done
-
Best wishes for 2022 everyone. Glad to see you here Max and to see what you became now, many congrats.
Thanks a lot to all the team that still works intensively every year to release better and stable versions of BMS that we love to fly on and that we grew with.
We owe you, the team, a lot, thanks again !
Fly safe -
@maxwaldorf Thanks for the update, if someone told me when F4 came out that I would still be running it over 20 years later (especially after the 1.08 release) I would never have believed it. Whilst i don’t get much gaming time anymore it still great to get the chance to occassionally have a quick mission. It has come a very long way over the years (and drove most of my PC upgrades and helped empty my bank account accordingly when I played games regularly ) I look forward to whats coming next!
Happy New Year to all.
Opener
-
@maxwaldorf
Thank you ! You are real people - a legend.
We believe in you. -
@dee-jay said in 2021 A Falcon BMS Year to remember!:
@bowser
1 > - AI fuel consumption is way off compared to human, especially low- and high level.
2 > - AI should not use afterburner when rejoining in flight.1 - Please elaborate.
How did you evaluated this?
Have you considered AI (robot) behavior and leader abilities to be a good leader considering the nature it his wingmen?2 - Please provide factual examples.
Also, consider that is could not be possible to make a difference between rejoin and other flight phases … so … if we prevent IA from using afterburner, it may be all also prevent them from using is in combat …
See the problem ?-
Straight and level flight with AI wingmen in formation, throttle constant just below mil power. Try at 1000 ft and at 35k feet. Evaluate fuel diference between ownship and AI.
-
Just get the AI separated from ownship and have them rejoin.
In 4.33 wingman did not use afterburner for rejoin.
-
-
@mav-jp said in 2021 A Falcon BMS Year to remember!:
@bowser said in 2021 A Falcon BMS Year to remember!:
- AIM-120 behaviour is unrealistic. RCS setting in cockpit determines exactly at what distance from the target the missile goes pitbull and that distance does not depend on actual RCS at all. The RCS selection in the pit should only provide an indication to the pilot about when the computer in the jet thinks the missile is going pitbull depending on the expected size of the target.
sorry but we do not understand what you mean ,
According to our information the way we modeled the RCS setting in SMS is realistic
the RCS option in the SMS determines at what range the Seeker will go HPRF / MPRF, independantly of course of the real RCS of the opponent
it also determines the size of the signals that the missiles radar will be looking for
we double check and the implementation in BMS is correctly done
Cannot find the documentation where I’ve read this. Thought it was in one of the MLU tapes, but there it’s actually not explained in detail. However, it might indeed determine the distance when the seeker goes active but this does not necessarily mean that it immediately acquires the target. For example a missile shot with RCS set to LARGE against a target with a very small cross section. I would expect that the missile would not acquire the target immediately when going pitbull, however in BMS the target ís always acquired directly when going pitbull (independant of actual target RCS and selected RCS for the missile).
-
@bowser said in 2021 A Falcon BMS Year to remember!:
@mav-jp said in 2021 A Falcon BMS Year to remember!:
@bowser said in 2021 A Falcon BMS Year to remember!:
- AIM-120 behaviour is unrealistic. RCS setting in cockpit determines exactly at what distance from the target the missile goes pitbull and that distance does not depend on actual RCS at all. The RCS selection in the pit should only provide an indication to the pilot about when the computer in the jet thinks the missile is going pitbull depending on the expected size of the target.
sorry but we do not understand what you mean ,
According to our information the way we modeled the RCS setting in SMS is realistic
the RCS option in the SMS determines at what range the Seeker will go HPRF / MPRF, independantly of course of the real RCS of the opponent
it also determines the size of the signals that the missiles radar will be looking for
we double check and the implementation in BMS is correctly done
Cannot find the documentation where I’ve read this. Thought it was in one of the MLU tapes, but there it’s actually not explained in detail. However, it might indeed determine the distance when the seeker goes active but this does not necessarily mean that it immediately acquires the target. For example a missile shot with RCS set to LARGE against a target with a very small cross section. I would expect that the missile would not acquire the target immediately when going pitbull, however in BMS the target ís always acquired directly when going pitbull (independant of actual target RCS and selected RCS for the missile).
The acquisition process is an entire other subject
4.36 is the answer
-
@bowser said in 2021 A Falcon BMS Year to remember!:
@dee-jay said in 2021 A Falcon BMS Year to remember!:
@bowser
1 > - AI fuel consumption is way off compared to human, especially low- and high level.
2 > - AI should not use afterburner when rejoining in flight.1 - Please elaborate.
How did you evaluated this?
Have you considered AI (robot) behavior and leader abilities to be a good leader considering the nature it his wingmen?2 - Please provide factual examples.
Also, consider that is could not be possible to make a difference between rejoin and other flight phases … so … if we prevent IA from using afterburner, it may be all also prevent them from using is in combat …
See the problem ?-
Straight and level flight with AI wingmen in formation, throttle constant just below mil power. Try at 1000 ft and at 35k feet. Evaluate fuel diference between ownship and AI.
-
Just get the AI separated from ownship and have them rejoin.
In 4.33 wingman did not use afterburner for rejoin.
Please read my explanations above
-
-
@mav-jp said in 2021 A Falcon BMS Year to remember!:
@bowser said in 2021 A Falcon BMS Year to remember!:
@dee-jay said in 2021 A Falcon BMS Year to remember!:
@bowser
1 > - AI fuel consumption is way off compared to human, especially low- and high level.
2 > - AI should not use afterburner when rejoining in flight.1 - Please elaborate.
How did you evaluated this?
Have you considered AI (robot) behavior and leader abilities to be a good leader considering the nature it his wingmen?2 - Please provide factual examples.
Also, consider that is could not be possible to make a difference between rejoin and other flight phases … so … if we prevent IA from using afterburner, it may be all also prevent them from using is in combat …
See the problem ?-
Straight and level flight with AI wingmen in formation, throttle constant just below mil power. Try at 1000 ft and at 35k feet. Evaluate fuel diference between ownship and AI.
-
Just get the AI separated from ownship and have them rejoin.
In 4.33 wingman did not use afterburner for rejoin.
Please read my explanations above
-
I don’t look at the fuel flow rate, I ask what the AI’s fuel state is before starting the run, and after. Fly for a few hundred miles and compare. Perhaps it’s also drag related, but did not test that.
-
So first wingmen were able to use AB for going defensive and not for rejoining but not anymore? Too bad if it cannot be fixed. In that case I would prefer the 4.33 behaviour.
-
-
@mav-jp said in 2021 A Falcon BMS Year to remember!:
@bowser said in 2021 A Falcon BMS Year to remember!:
- AI fuel consumption is way off compared to human, especially low- and high level.
AI fuel consuption is off indeed (i wouldnt say WAY off ) , but actually the AI burns less fuel than the player , didyou think it was the opposite ?
Mav , I think you’re right and it’s not a bug. I believe the AI just thinks more about completing it’s task then saving fuel. For example, on the carrier when in tension on the Cat, the AI is in burner literally for minutes before launch. Also, I’ve called my wingie go spread and there goes the 'burner. If I do a fast climb from launch AI is in burner even when I’m not.
The good news is you can compensate quite a bit with procedure(ie:being a good Lead). I’m doing some testing with launch procedure( such as taxiing close to launch time). Also, I’ve taken to doing a slower climb out, at the optimum climb speed for your jet. Things like that. -
-
2021 was a Falcon BMS year to remember for sure. I have the impression though that 2022 will be also an unforgetable year!
Thank you all! -
A great time to always remember
-
Hi!
@bowser said in 2021 A Falcon BMS Year to remember!:
@dee-jay said in 2021 A Falcon BMS Year to remember!:
@bowser
1 > - AI fuel consumption is way off compared to human, especially low- and high level.
2 > - AI should not use afterburner when rejoining in flight.1 - Please elaborate.
How did you evaluated this?
Have you considered AI (robot) behavior and leader abilities to be a good leader considering the nature it his wingmen?2 - Please provide factual examples.
Also, consider that is could not be possible to make a difference between rejoin and other flight phases … so … if we prevent IA from using afterburner, it may be all also prevent them from using is in combat …
See the problem ?-
Straight and level flight with AI wingmen in formation, throttle constant just below mil power. Try at 1000 ft and at 35k feet. Evaluate fuel diference between ownship and AI.
-
Just get the AI separated from ownship and have them rejoin.
In 4.33 wingman did not use afterburner for rejoin.
1 - What do you consider as “lots of fuel / big difference” on an entire flight? 1000lbs, 2000lbs, 3000lbs ? … We may not have the same definition of “big difference”.
I will try to check is (again) but I didn’t had big difference since low time ago (of course, on a standard navigation without defensive/offensive actions).
The best would be to join an ACMI if you can and the TE so I can fly it also on my side in the exact same conditions.
2 - The issue was that the AIs were in some case NEVER able to rejoin the flight after a takeoff (I.E. runway 36 departing northbound on a straight track).
So we made some leader actions to maintain 6000kts outbound and slower speed (not sure about the speed) until #4 rejoin (that will be allowed to use a bit AB for this and should not take long time) … then, once rejoined, leader is allowed to start climb again and once cruise FL reached, accelerate to match the TOS.In case of human leader, the rejoin plan can even be smarter (making a triangle to help the wingmen to rejoin) that ca can’t do for AIs. Also other things about flight planning but none ensure 100% efficiency. The only way to fix this was to re-allow IA to use AB for rejoin for a shorter period of time (if human leader is not dumb and take care of his wingmen).
Ideally, the AI leader would be amble to perform a pre-planed trajectory routine of few turns (one or two) to help the rejoin,
… examples:
… but it is not possible at that point to hardcode it. It is always a matter of difficult compromises.
-
-
-
Tested it again, seems OK now with AI burning a little less indeed. Think my latest test stems from 4.34.
-
Just to be sure, I mean rejoin during any flight phase (for example after air combat, especially with the new AI tactics, wingmen like to get separated and use plenty of burner to catch up again). But I assume that what you mention about take-off is also valid during mid-flight. For pure AI flights is it not possible to have the leader maintain a relative slow speed until its wingmen have rejoined after completion of the task that caused the separation? And for after take-off, currently the flight plans have a 6 min holding, doesn’t that solve the rejoin after take-off?
-