Auto-ATC... what does it actually offer ?

[Bjoern 1,157]

17 hours ago, mSparks said:

[Documentation]

Not my fault!

Aviation comms documentation is basically nothing more than a ton of multiple choice questions along the lines of "Q:An aircraft will be in the best range for VFR communications if it is: A. in the vicinity of the airfield." -> of practical use to exactly no one who ever sat in a cockpit. I had to start completely from scratch. The latest phone app has a "ground school" section, there is still a lot more to go in that, and I'm already working with or in discussion with a couple of ATOs to get it right, everything basic needed for startup->taxi->take off->navigate->approach->land->taxi will be [finally] written up and reviewed very soon.

[Airports]

Airports are divided into three levels internally,

lvl0->no taxi lanes defined, these you will just be told to hold short of the runway (quite a significant number of airports, and the hardest to get ai plane taxi routing right, which often causes the ai planes to fail to initialise)

lvl1->defined and named taxi lanes, these you will get taxi instructions which must be read back correctly to progress, ai planes should behave well, a few thousand airports. 

lvl2->hand crafted additional information (such as disabled runways for take off land) and additional testing (currently only EGLL, EGCC and KSEA)

More details on which airports these are will go up on github once recompiling them all to the latest standards is complete, a lot of the current strange behavior is now fixed bugs related to building the airport maps.

[GA]

To a fashion, there is actually GA traffic, and they can fly flight plans, such as the S76 I create the custom flight plan for in the video above, or the TUCA on the youtube channel. The problem is currently you'll most likely never see them default because they all fly out of either untowered airports which the AI doesn't support yet, or small lvl0 airports that they fail on. Fixing that is planned for soon after the ground school basics are in.

Re: Documentation
Have said it before, will say it again: You're writing the recognition and reply scripts for the controllers, so transcribing those to provide a starting point is an option and makes things easier for newbies.
Or at least try to get users to practice all available requests and readbacks in ground scholl, including that sample flight (after you're done implementing comms for the respective flight rules into STT and TTS).

Re: Airports
MSFS handles airports without defined taxi nets by spawning departing AI on the runway and despawning it after rolling out on the runway after landing. Looks ugly, but at least it opens the entire selection of airports to AI traffic, regardless of development state. If the user adds an updated taxi net, the AI will automatically start to use it.

Re: GA
Ideally, you'd have GA traffic specific to a certain airport. Not sure if I've said this before, but I have a worldwide list of airports with more than 1 million pax p.a. which I've used as a blacklist for very small GA (C172 or so) while generating global GA AI traffic for FSX. (FSX' traffic engine isn't very good at squeezing small GA between airliners on arrival.) You could use this list to broaden the list of supported airports before going "all out" with global generation later. Holla if you want it.

 

7 hours ago, mSparks said:

I went with the samsung A70 for my latest phone.

I've seriously looked at this one when looking for a new phone last year, but spent the extra money and went for a Pixel 3a instead. It turned out to be a perfect replacement for my Lumia 950. Except for lacking Win Mobile's sleek UI and having more data mining features than MS could ever dream of, of course.

 

2 hours ago, mSparks said:

I meant to comment on this before. still not 100% exactly what this will look like, technically its the automatic cross country frequency handovers that aren't implemented, of which areas served by center is one of many options (VFR seems to get handed over from one airport to another for example), still working on a basic set of rules that will simulate that with reasonable realism. If any body has seen a guide on which and how its decided to hand an aircraft over to another frequency while it flies cross country would be very glad to read it.

Its not as simple as ->entering new class A->hand over to class A center. For a start VFR pilots never enter class A airspace, still need handing over.

Airspace data can be obtained from OpenAIP, but you might need a lot of post-processing to make it work with Auto-ATC.

https://www.openaip.net/

For ARTCC/ACC airspaces, I can offer the 2005ish data from MSFS (with a custom frequency mapping based on what Google gave me plus a lot of imaginative ones if a search turned up empty). I've imported it into QGIS (files are avialable) and exported a simplified version that splits up the borders by XP's 1 by 1 degree terrain grid. Using that one, you'd have to do some more calculations to determine the "owner" of the current and adjacent terrain tile to account for multiple sectors sharing the same tile, but my Lua based code seemed to work and I'd put that into the package.
MSFS by the way even subdivides the airspace polygons for different controller frequencies for the same ACC/ARTCC, but this sometimes tended to produce overly frequent controller handoffs.
If you want up to date data, the only sources are the ICAO for $5000 or the IVAO airspace file without frequencies and potential border simplification.

Edited January 20, 2020 by Bjoern

[mSparks 2,090]

29 minutes ago, Bjoern said:

ou're writing the recognition and reply scripts for the controllers, so transcribing those to provide a starting point is an option and makes things easier for newbies.
Or at least try to get users to practice all available requests and readbacks in ground scholl, including that sample flight (after you're done implementing comms for the respective flight rules into STT and TTS).

Yes, exactly this, working with the ATOs on all of that.

30 minutes ago, Bjoern said:

Re: GA
Ideally, you'd have GA traffic specific to a certain airport. Not sure if I've said this before, but I have a worldwide list of airports with more than 1 million pax p.a. which I've used as a blacklist for very small GA (C172 or so) while generating global GA AI traffic for FSX.

The ones I am using are real life filed flight plans, a sample of British airways planes looks like

{"dst":"ENGM","src":"EKAH","af":"J328","callsign":"speedbird","af_iata":"FRJ"}
{"dst":"EGLC","src":"EGPD","af":"SB20","callsign":"speedbird","af_iata":"S20"}
{"dst":"EGLC","src":"EGPD","af":"E170","callsign":"speedbird","af_iata":"E70"}
{"dst":"EGLL","src":"EGPD","af":"A321","callsign":"speedbird","af_iata":"321"}
{"dst":"EGLL","src":"EGPD","af":"A320","callsign":"speedbird","af_iata":"320"}
{"dst":"EGLL","src":"EGPD","af":"A319","callsign":"speedbird","af_iata":"319"}

activating AI aircraft at an airport near a user randomly selects some of those flightplans and spawns AI aircraft, for example, activating EGLL from that list might make a E170 with a speedbird callsign that will fly to EGPD.

There are similar flight plans for 527 different C172s in the data I have (e.g. CGPIQ from CYKF), looking at the data currently non of the airports they fly from have passed basic testing.

lvl0 airports aren't necessary small, lvl1 aren't necessary large, just haven't had the loving of lvl1. so all I have to work with is for example

rather than nice point to point navigation from apron/gate to runway, as of right now more lvl0 are now compiling without problems - need to add building avoidance for 3d airports. lvl1's should be rebuilt by tommorrow and live after a few days of unit testing.

making an airport lvl1 requires adding the taxiway data to the airport in wed and exporting it to the gateway.

58 minutes ago, Bjoern said:

Airspace data can be obtained from OpenAIP, but you might need a lot of post-processing to make it work with Auto-ATC.

airspace class doesn't help the way I was expecting it to, it would be live already if it did. There are multiple center frequencies within a single airspace, and for example, HF is used for transatlantic flights.

London ACC alone uses

118.480 
118.825 
119.725 
119.775 
120.025 
120.175 
120.475 
120.525 
121.025 
121.225  
121.325 

or 124.6 for FIS east -> "somehow". I'll probably just end up with some semi sensible rules on when and how to trigger a transfer, probably based on distance from the transmitter and number of traffic on frequency (will also include things like kicking everyone off frequency when someone declares an emergency). I'm also missing that data for the US atm.

Getting there though. would really of preferred to be a bit further on than I am, but stability and performance have taken a front seat. The only todo currently left there is fixing the compiler for lvl0 airports.

 

[jcomm 5,960]

Well,

all I wish is that you can keep your mind well feed by these excellent ideas for improving Auto-ATC.

For those users not yet familiarized with this App I suggest looking in youtube for a few rather informative videos.

Only thing missing on my side to adopt it is a smartphone 🙂  I need to buy one ....

I can't count as a "connoisseur" in as far as this type of applications go, but I have the feel this one in particular might well conquer my attention and dedication !

Keep the good work Sparks !

Suggestion / Hint: Appart from these "hobby", you're a rw "rotary wing driver" too, so, I would say, the type of element that should fit perfectly in the LR team, to help X-Plane continue to grow towards an even more complete flight simulation platform. Have you contacted Austin ? Why not offering him availability to help with an ATC solution for.., say... XP12 ?

And..., from the various posts I folowed at the .Org where you offer some really interesting comments regarding the rotary wing model in X-Plane from the pov of someone who does it for real too, and loves that particular type of aircraft, I am sure the team could also gain from it.

Edited January 21, 2020 by jcomm
typos

[Bjoern 1,157]

17 hours ago, mSparks said:

airspace class doesn't help the way I was expecting it to, it would be live already if it did. There are multiple center frequencies within a single airspace, and for example, HF is used for transatlantic flights.

London ACC alone uses

118.480 
118.825 
119.725 
119.775 
120.025 
120.175 
120.475 
120.525 
121.025 
121.225  
121.325 

or 124.6 for FIS east -> "somehow". I'll probably just end up with some semi sensible rules on when and how to trigger a transfer, probably based on distance from the transmitter and number of traffic on frequency (will also include things like kicking everyone off frequency when someone declares an emergency). I'm also missing that data for the US atm.

But how are you triggering transfers from, say, London Center to Amsterdam Center?
If tile queries and geodata isn't your thing, you can also imitate 124th ATC's ACC/ARTCC assignment logic, which has a database of all (relevant) airports in each ACC's area of responsibility. Picking the currently responsible ACC is done by querying the nearest airport and trying to match it with one from the database. A match triggers a handover if not already checked in with the ACC.
Flying the Atlantic or Pacific might require workarounds as the system does not account for it, but it works reasonably well over land if the basic database is comprehensive enough.

Edited January 21, 2020 by Bjoern

[mSparks 2,090]

10 minutes ago, Bjoern said:

But how are you triggering transfers from, say, London Center to Amsterdam Center?

reception is based on the transmitter locations, the list of which update as you fly, my general plan is to add "contact ZZZZ on xxx.xx" nodes to the detailed flight plan when it is constructed that would then function more or less identically to the "fly heading" commands that get issued in experimental IFR. just not sure exactly that will look like - needs to be before you are out of range of the current station, and if possible when you are in range of the next one. Which is a long way from always achievable (in which case it needs to be "contact ZZZZ on xxx.xx when able"). almost certainly before you actually enter the zone.

At mach 0.86 at 35,000 feet you are into and out of range of even strong transmitters every 20 or so minutes (11 miles a minute, ~230 miles range) - and that's best case if you fly directly overhead; half that range at a FL110 without oxygen.

global terrain system is in there already too, but at the moment its just used for AI flying flight plans to keep them always no less than 1000' AGL. (about 90% tested locally)

Todays ground school drafting was actual some of the more gritty details of this -> such as flying IFR in weather when your radios die.

7 hours ago, jcomm said:

all I wish is that you can keep your mind well feed by these excellent ideas for improving Auto-ATC.

I hope I can get it to meet my expectations, or I'll need to go all fight club on my hind, last 12 months have been.... tough... Mrs has been amazingly understanding, but even she is definitely starting to miss our usual lifestyle now...

7 hours ago, jcomm said:

I am sure the team could also gain from it.

I have thought about it a few times (especially when their job offers have gone up), but even now they don't need to pay me, and I definitely don't need any other obligations than the ones I already have. Always look forward to what the future may hold.

[Bjoern 1,157]

20 hours ago, mSparks said:

reception is based on the transmitter locations, the list of which update as you fly, my general plan is to add "contact ZZZZ on xxx.xx" nodes to the detailed flight plan when it is constructed that would then function more or less identically to the "fly heading" commands that get issued in experimental IFR. just not sure exactly that will look like - needs to be before you are out of range of the current station, and if possible when you are in range of the next one. Which is a long way from always achievable (in which case it needs to be "contact ZZZZ on xxx.xx when able"). almost certainly before you actually enter the zone.

What about flights without a flight plan? Like when you're going up for a hop near FIR boundaries, want information services and then cross from one FIR into another?

[mSparks 2,090]

4 hours ago, Bjoern said:

What about flights without a flight plan? Like when you're going up for a hop near FIR boundaries, want information services and then cross from one FIR into another?

no flight plan VFR its the PIC's responsibility to contact the station in control of an airspace or service. dial in a frequency and make a request. If you are out of range of the transmitter calling you (or on a different frequency), you just dont get it

Calling a position->next can add a contact node the same way it will for the flight plan.

"go ahead" in place of "say again" with no request hit is already queued for the next update.

Edited January 22, 2020 by mSparks

[mSparks 2,090]

43 minutes ago, stans said:

I can't imagine that there would not be some sort of flight planner.  As for AI ATC, I sure hope it is smarter and more adaptable than what was in previous versions of MS's flight simulator series.  I would like to be able to request course deviations so as to fly around thunderstorms and to be able to declare emergencies.  I also hope that a new AI ATC would be smarter and not make you change course, turning 90 degrees left, then 180 degrees right, then 90 degrees left putting you back on the same track that you were on.  That is asking a whole lot, I know.

Quoted from another thread but more relevant here.

Really good suggestion, So far I've bounced "unable to comply" down the to do list a few times because the only example I had come across was "unable to comply due weight" which didn't seem that important, "unable to comply due weather request heading/altitude" is definitely one for the top of the list.

great progress on basic ground school, cross country frequency handover logic (at least outside of the US) and general language this week, still a fair bit more to do. 

I did "declare an emergency" almost as a joke quite a way back just for testing some sequence stuff

New airport builds finally succeeded yesterday and testing is looking good (3,772 lvl1 airports), had one moment where I thought it was still broken, but LFPG really is just crazy complex

[mSparks 2,090]

Think I'll "cheat" a little on centre frequencies first and just use a "looks OK" frequency rather than try and estimate more precisely what would actually be used. Flight testing next Experimental IFR update next week.

[jcomm 5,960]

Looking ( actually it's - sounding ) cool!

[mSparks 2,090]

22 hours ago, jcomm said:

Looking ( actually it's - sounding ) cool!

Looks too 🙂

S76 (short callsign Golf Kilo Golf) takes off at 1min 40

C172 (short callsign Golf November Alpha) takes off at 6min 10

That video just needs the plugin (and all still free atm) - its not using the phone. I started prepping for an experimental IFR (EIFR) flight from EGCB to EGCC and found a load of bugs that were stopping the GA I expected at EGCB spawning. Needs these updates fully testing and pushing to the AI server before they are live, expecting that this week assuming nothing to demanding is found during the next lot of EIFR testing.

 

Edited February 18, 2020 by mSparks

[jcomm 5,960]

8 hours ago, mSparks said:

Looks too 🙂

S76 (short callsign Golf Kilo Golf) takes off at 1min 40

C172 (short callsign Golf November Alpha) takes off at 6min 10

That video just needs the plugin (and all still free atm) - its not using the phone. I started prepping for an experimental IFR (EIFR) flight from EGCB to EGCC and found a load of bugs that were stopping the GA I expected at EGCB spawning. Needs these updates fully testing and pushing to the AI server before they are live, expecting that this week assuming nothing to demanding is found during the next lot of EIFR testing.

 

Also take care with whatever the 11.50 update, about to rollout, can modify in X-Plane's kernel ...

[mSparks 2,090]

9 hours ago, jcomm said:

Also take care with whatever the 11.50 update, about to rollout, can modify in X-Plane's kernel ...

should be 11.50 ready. Just not sure the rest of the world is. Quite the stir coming when that monster drops.

[jcomm 5,960]

It's an year since my initial post on this subject - Auto - ATC - and I thought I might bring it back to the top of the Forum 🙂

I haven't tested this software, mostly because I do not have the time or PC availability / sound conditions at home to play with it, or actually to do any "pro-like" playing with flight simulattion, but I am sure I will enjoy it & when I find the time and conditions to set it on my desktop 🙂 and profit the most out of what it offers.

Also until about a month ago I didn't have a smartphone... I now have my first smartphone ( Android ), and while still trying to learn how to use this fancy gadget, I already make some good use of soime Apps for it.

Auto-ATC is the example of a very useful tool for those starting in aviation, and for me it will for sure be useful because my only "real life ATC complexities" were experienced in simulators ( in the gliders we avoid it at all cost, and actually sometimes prefer not to even be noticed .... Shhhhh ! )

Lookig fwd into a full IFR version, with Center, Approach, and even Oceanic stuff 🙂

Unfortunately I've heard mSparks is porting it to Windows only, and will support MFS as the single platform 😕 ( hehehe, joke of the day !!! ) 😎

Edited January 7, 2021 by jcomm

[mSparks 2,090]

7 hours ago, Bjoern said:

I was especially referring to the app whose changelogs were a bit on the barely informative side in the past.

Majority of the app updates have been fixes to auto crash reports or layout fixes for different devices (thought it better to reply here)

Only really new thing that I haven't especially documented is 

But I don't think anyone's really noticed yet, they're all still mostly learning how to fly it (the 744). I'll probably add in support for others reasonably soon.

Also been using git issues for the app, quite a few of the big tickets have been quietly signed off over the months.

15 hours ago, jcomm said:

Unfortunately I've heard mSparks is porting it to Windows only, and will support MFS as the single platform 😕 ( hehehe, joke of the day !!! ) 😎

I expect to come back to MFS once MFS is out of alpha 😁, iPhone is higher on the agenda for now. 

15 hours ago, jcomm said:

Lookig fwd into a full IFR version, with Center, Approach, and even Oceanic stuff 🙂

I had lots of meetings planned for last year to discuss doing all that properly that never came to fruition because of the zombie apocalypse, hopefully they can happen this year, I'm most excited to finish up whats left of the language model, been talking/thinking about a tool to view/edit that for most of the last 12 months and finally got that working a few hours ago, still riding high from seeing it work and trying process in my head "what next". 

Bjoern lambastes me for not documenting it more but I'm really working to the philosophy that its

"a button you press to transmit on a frequency that should behave the same to pressing transmit on a real radio - and everything else is fluff"

about done with the fluff now, onto finishing the meat.