Heads up on new AI separation utilities for P3Dv4

[vincentrouleau 71]

FYI, I had 2 CTD in a row in P3dV5.1hf1 at EGCC leaving gate 206 with a AiCull  ( running not executing) crash in the event viewer:

Restarted without AiCull and I am ok now.

Faulting application name: AICull.exe, version: 1.0.1.1, time stamp: 0x5f03bd07
Faulting module name: ntdll.dll, version: 10.0.19041.662, time stamp: 0x27bfa5f0
Exception code: 0xc0000005
Fault offset: 0x0000000000045f86
Faulting process id: 0x4cfc
Faulting application start time: 0x01d6d0b9645b315b
Faulting application path: C:\Users\vince\Documents\Prepar3D v5 Add-ons\AICull\AICull.exe
Faulting module path: C:\WINDOWS\SYSTEM32\ntdll.dll
Report Id: 0f109dff-e02e-445a-893d-fc0c64e71fc7
Faulting package full name: 
Faulting package-relative application ID: 

Edited December 12, 2020 by vincentrouleau

[Clipper Ocean Spray 189]

On 12/3/2020 at 4:16 PM, colinosullivan said:

Hello all,

The first issue I would like to address is the amount of holding being created. It seems that there may be in error in the spacing between aircraft at times? Not entirely sure. Reference " Holds ". I do have a pretty massive amount of aircraft in my sim, so issuance of holds and updating them could be a problem, however it looks like they keep being put into them.

Secondly, reference " Holds 2" . In this image we can see an aircraft inbound to LAX going into a hold for no apparent reason. With this being said, my separation factor is 2 nm. 

Holds - Link to image https://photos.app.goo.gl/ze8thLB2CXVTptHRA

Holds 2 - Link to image https://photos.app.goo.gl/zG61LH4387TfGFt96

An AI is put on hold based on an estimate of how close the AI is to another AI arriving for the same runway, not AI immediately ahead/behind.  In other words, an AI approaching from the north 20 nm out may be put on hold for an AI approaching from the south 20 nm out because they are both calculated to arrive at the runway within the separation distance and even though they are currently 40 nm apart.  Having said that, the distance calculation are estimates only especially since they have to take into account the predicted flight path.  Additionally, if an AI on short final has to hold, all AI queued up have to hold in order to prevent an AI go-around bug.  Finally, if you're running high AI densities, I would recommend a max no. of holders in the config file because a high number of holders tends to beget more holders.  With a max number of holders, excessive holders (tending to happen in the first arrival wave after a situation is loaded) will egress will egress the reality bubble, but you should still get a pretty busy and organized landing flow.

 

On 12/12/2020 at 12:18 PM, Ray Proudfoot said:

A week ago I took off from Rwy25 at Rome LIRF. As I was on the take-off roll I noticed an Ai suspended in mid air having taken off on 16R. As soon as my wheels left the ground he continued flying. Is there any advantage to this? We were never in any danger of a collision.

On another flight I was a couple of miles out from Rwy 04 at EHAM when an Ai taxied onto the runway and Radar Contact issued a GA. The visibility was poor so I had no chance of seeing him so I could zap him with FSUIPC.

Sadly there is no log created with AiFlow so it’s difficult to analyse what it’s doing. Any chance a log could be generated?

No advantage.  Thanks for identifying that, I need to tweak the logic so it doesn't impact AI in takeoff2 but airborne.

Perhpas 1500 ft agl is too low of a trigger alt.  I should probably revisit that value and/or add a parameter to the config file to customize.

 

On 12/12/2020 at 3:20 PM, vincentrouleau said:

FYI, I had 2 CTD in a row in P3dV5.1hf1 at EGCC leaving gate 206 with a AiCull  ( running not executing) crash in the event viewer:

Restarted without AiCull and I am ok now.

Faulting application name: AICull.exe, version: 1.0.1.1, time stamp: 0x5f03bd07
Faulting module name: ntdll.dll, version: 10.0.19041.662, time stamp: 0x27bfa5f0
Exception code: 0xc0000005
Fault offset: 0x0000000000045f86
Faulting process id: 0x4cfc
Faulting application start time: 0x01d6d0b9645b315b
Faulting application path: C:\Users\vince\Documents\Prepar3D v5 Add-ons\AICull\AICull.exe
Faulting module path: C:\WINDOWS\SYSTEM32\ntdll.dll
Report Id: 0f109dff-e02e-445a-893d-fc0c64e71fc7
Faulting package full name: 
Faulting package-relative application ID: 

OK, will keep an eye on this.  About to upgrade to 5.1 now that VR seems to be sorted out for my setup.  If I do experience this, shouldn't be hard to trouble shoot as AICull is doing very little when not executing a cull.

[Michael Leung 5]

Awesome set of utilities, Roland! Bravo!

I love the improved pushback - is it possible to consider having AI pushback and turn 90 degrees instead of 180 degrees to point the opposite direction?
It seems like most aircraft IRL push back 180 degrees, stop and deploy flaps, detach the pushback tug, start engines, etc... it would make it so much more realistic.

Thanks for the consideration!

[DuduGuimaraes 10]

Hi guys

Does using aiground + aiflow have a big tool on frame rates ?

Thanks

Eduardo

[pmemery 4]

Eduardo, not in my modest configuration (6700K + GTX1060 6 Go)

[JoeFackel 760]

11 hours ago, DuduGuimaraes said:

Hi guys

Does using aiground + aiflow have a big tool on frame rates ?

Thanks

Eduardo

No, virtually none. The AI itself will steal some frames as it does always.

[Clipper Ocean Spray 189]

On 1/16/2021 at 4:55 PM, Michael Leung said:

Awesome set of utilities, Roland! Bravo!

I love the improved pushback - is it possible to consider having AI pushback and turn 90 degrees instead of 180 degrees to point the opposite direction?
It seems like most aircraft IRL push back 180 degrees, stop and deploy flaps, detach the pushback tug, start engines, etc... it would make it so much more realistic.

Thanks for the consideration!

Thanks!  Good suggestion re: 90 deg turns.  My prior AIController read the scenery .bgl files and knew the taxi routes out of the terminal area and thus could figure out the correct 90 deg. turn direction.  However, AIGround is sort of a bare-bones version of AIController and doesn't read the scenery bgl files, thus the 180 deg turn to ensure the AI doesn't end up accidentally pointing into the corner of a terminal area after pushback.  I might be able to figure out a way to reliably do 90 deg. turns without reading .bgl data though, I'll keep looking into it.

[Deltaair1212 53]

Hey all! I see this thread has been quite dead since the beginning of the year, so I apologize for digging the grave up but I am having some issues… well… rather annoyances with AI Flow and AI Ground. With Flow, I am experiencing aircraft suddenly climbing straight up in the air as I pass by them or if I get too close and they just go straight up, stall, and then fall down to the ground. In real life, flight levels are separated by 1000ft in RVSM airspace. To me, this program does not accurately simulate this like in real life. I’ll be at FL390 and a aircraft at FL380 going in the opposite direction will just randomly go straight up in the air. I went into the settings and even put 1 mile separation between AI and user aircraft but nothing works. 
 

with AI ground, I am experiencing AI traffic coming to a complete stop even though I am nowhere near them whether taxing or parked on the apron. AI traffic will still hold in position for my aircraft even when I’m parked and engines off and everything. The program doesn’t take into account that once your aircraft is off, you don’t need for AI to take your aircraft into consideration. Is it possible to put into a future update that anytime your parking brake is set that the AI aircraft disregards your plane? That would help out so much! Also it says that flashing your landing lights will tell AI to continue taxiing but this does not work all the time. Only about 20% of the time for me. I’ll flash my lights on and off and still won’t get them to move. So my AI just sit there literally for hours and do not move. 
 

sorry for the lengthy message but hopefully this gets to the developer. BTW, I am using P3Dv4.5 with AIG for my AI traffic, and ProATC/X for my ATC. 

Edited September 3, 2021 by Deltaair1212

[FKHUL 0]

I've never been able to get AI Ground to connect and with AI Flow it usually shows this message...

AI Flow Ver. 1.1.0 for P3D V4 or V5

Waiting to detect AI aircraft within 55556 meters (29.9947 nautical miles) of the user's aircraft's current position.

Traffic flow operations may not be immediately reported (e.g., AI not currently close to each other, light traffic).

The AI Flow Program has connected to the P3D flight simulator!

\ User's aircraft is nearest ????.  ||Closest ICAO will update as user aircraft moves through different areas  ||

Any ideas what the problem could be? I really would love to get these to work to see the amazing functionality but have no idea why AI Flow would only work sporadically at a few airports and why AI Ground wouldn't work ever.

Thanks!

Jeremy

[Wanthuyr Filho 83]

These are great tools, thank you!

Whenever flying at cruise level and another plane passes nearby it acts erratically, like it zigzags up and down in an abrupt manner. Even the contrails are drawn erratically too. Is it possible to fix this behaviour?

[Papacoach 75]

6 hours ago, Wanthuyr Filho said:

Whenever flying at cruise level and another plane passes nearby it acts erratically, like it zigzags up and down in an abrupt manner. Even the contrails are drawn erratically too. Is it possible to fix this behaviour?

Try this setting...  USERAISEPARATIONDISTANCE=0.0

[freddy_ 9]

This is an older thread, but I don't see anywhere else to post this question.  (I can always email AIController2 as per the documentation, but I will try here first).

I am using P3D 4.5 HF3 with AI from an older traffic program (Traffic X). The AI seemed fine enough to me without using AIFlow and AIGround ... however, upon discovering these utilities last week I did think about the number of go arounds I see, along with short takeoff and landing rolls, etc.  So these utilities sounded great and I installed them.

I am having a problem with AIGround (Rev 1.1).  With the AIPushBackHoldTime parameter, aircraft tend to "freeze" dead on the spot. It is a noticible freeze.  For prop aircraft the propellers no longer animate, and the aircraft looks like it is stuck in time.  This is extremely noticible with GA aircraft and aircraft like Dash 8s and ATRs. Prop aircraft have a nice looking, realistic, slowly rotating stroboscopic wagon wheel animated effect for their propellers. This is shown as soon as engines start, and they begin their pushback. But, when the AIPushBackHoldTime kicks in, the aicraft "stop", dead, and their propellers freeze (they still show a see-through "rotation effect", but it is completeley static ... it is no longer the realistic, slowly rotating stroboscopic wagon wheel effect).  As the aircraft are released from the AIPushBackHoldTime parameter, the realistic, slowly rotating stroboscopic wagon wheel effect returns.  This whole spinning prop, push-back, freeze, no spinning prop, unfreeze, spinning prop returns sequence looks extremely unrealistic.

It made sense to me to set AIPushBackHoldTime=0 to turn this parameter off.  The DOS based logging screen shows that it accepts this parameter value and it has been set to "0".  However, as the very next aircraft begins it's pushback, the AIGround program closes.  I assume it has crashed, but it closes in a nicely handled way (a credit to the way the program has been coded).  The Windows Event log does show an AppCrash, but there really isn't much detail in there that I can use (although I can post it here if needed).

Based on this, I tried to set the parameter to "0.01", but the DOS based logging screen showed an error for this parameter value and advised that this likely because an integer is expected and a decimal has been used (again, that's great coding and well handled).

So, I set "1" second for the AIPushBackHoldTime parameter. With the parameter value set to "1", the AI now pushback, commence their turn toward the taxiway and don't appear to freeze.  Things look OK.  But I am worried that, based on the programmed randomness of this parameter, (it will randomly choose between 1 and 2 seconds for the hold, using a setting of "1" second) whether some aircraft may end up doing a sudden, jerky-looking, 1 to 2 second "freeze, stop and stutter" kind of thing?

Has anybody else attempted to turn this parameter off using "0" and experienced a similar program close/crash?

I am hoping that maybe Clipper Ocean Spray still reads/follows this thread and might respond.

In summary, I am wondering ...

• if setting "0" for this parameter, to turn it off, is a bug (as yet unreported?) and a quick hotfix could be done, or
• if the setting of "1" is the equivalent of turning the parameter off, or
• if "1" is not the equivalent of turning it off, whether there's another workaround that can be suggested.

Edited December 13, 2021 by freddy_