Issues with AI and the ASR mod - I am dumbfounded

[Booth.the.doberman]

I am using the ASR 1991 F1 cars and running a 1991 season. I usually spend all week learning, training and tuning with Friday nights being race night. Tonight was Monaco with the RaceRFactor Monaco track. I have 22 cars running a combination of the Ferrari, Benetton, McLaren and Williams. I finished first. Congratulations, right? I was the only car left on the track after the 28-lap race ended. On lap 1, six cars retired. Four of them were fuel, one accident and one simple DNF. On the second lap, three accidents and a fuel. By lap 18, the fourth place card expired. In the end there were 7 fuels, 10 accidents, 2 engine, 2 DNF. Okay, now I am pretty bad, so I have the AI set really low as I learn this sim racing thing so I can be competitive (I qualified 15th so I'm not running away with the races). But I just don't understand the fuel issue...especially on the first lap. A whole bunch of them went into the pits. Why didn't they refuel? I think they went in when the safety car was out; I remember reading something about pitting when the SC was out.

I should add this: I started the race twice. The first time, a ton of cars left the track on the first lap (the SC was out again because of a crash). I thought there was something wrong so I restarted the race (the results of the second start is what I describe above where I was the only one to finish). Anyway, I looked at the LogAnal report on that start and it shows that 12 of the AI cars retired for fuel on laps one and two.

I am dumbfounded. Anyone have a wild guess what the hell happened here?

 

[Booth.the.doberman]

I just finished a 28 lap race at RaceRfactor-Monaco using the ASR 1991 F1 mod against a field of 24 AI. Nearly all 24 AI finished my race except for: 2 AI: fuel (lap1) and 1 AI retired with suspension problems!

With my results it doesn't appear to be an issue with the ASR 1991 mod or RaceRfactor-Monaco

Well, that just adds to the confusion.

Just to be sure: you were running the 1991 cars? As far as I know, there isn't a 1991 mod. There's a 1992 mod (ASR_Formula_OWC_92_186.rfcmp).

For my 1991 season, I downloaded the four separate 1991 cars individually from the ASR website. I used one of each of those as a "master" to skin 21 AI cars, create new drivers and generate the AI all in the "Tuning" menu.

The 1991 cars are version 1.8 except for the Benetton which is 0.5. I created a race event using those cars and the tracks that are available to correlate with the 1991 season (there is no Phoenix and most of the other tracks are from the 2000s, but, hey, that's the best I could do).

Not only did I have a lot of AI retire because of fuel, including on lap 1, but I had a lot of accidents (none of which involved me, by the way) AND I had all AI retire AND I had that aberration with the number of laps in the race where I scheduled it for 20 and 28 were actually run. I can only think that the laps with the SC on the track were not counted against the 20 laps. As a matter of fact, Spotter 2.0 kept telling me over and over on each lap the SC was out that there were the same number of laps to go. The SC came out on lap one, for example. When I crossed the start/finish line the spotter said there were 20 laps to go. When I crossed it the next time, he said there were 20 laps to go. On the third, he said there were 20 laps to go. Something is really hosed up here. Steam's validity checker says everything is just fine and dandy.

By the way, ASR posted an update for one of the two 1992 cars tonight and said updates for other 1992 cars are coming soon.

 

[Ace King]

Well, that just adds to the confusion.

Just to be sure: you were running the 1991 cars? As far as I know, there isn't a 1991 mod.

The 1991 cars are version 1.8 except for the Benetton which is 0.5. I created a race event using those cars and the tracks that are available to correlate with the 1991 season

Yes, i had to download each 1991 car v1.8 and Benetton v0.5.

Note: Did you "force setups" so all AI use same setup configuration? I forced myself and AI to use "early version setup".

You stated that you used one of each of those as a "master" to skin 21 AI cars, create new drivers and generate the AI all in the "Tuning" menu.

I believe it's possible that when you generated the AI in the tuning menu that the "AI settings" might not have been ideal thus causing some type of issue! Note: I didn't create any new AI drivers in the tuning menu!!

Hope this helps, Good luck

 

[Booth.the.doberman]

With the McLarens, I had to specify one of the forced setups or the new skins wouldn't appear. I didn't force it on any of the others because they all seemed to skin properly. I suppose the generated AI files could be less than ideal. None of them looked out of the ordinary and I didn't see any of the issues on Road America or Sao Paulo, but I think those had the previous versions of the cars. I guess a good test would be to load up all the 1991 cars (originals plus my add-ons) and run again at Monaco and see what happens. Awfully thick field for a street circuit but it will at least give me some idea about the fuel problem.

 

[Ace King]

I would do the same test as i did by only loading the 1991 cars and don't use your add-ons. Just run the 1991 cars "stock" without those generated AI talent files as this might be a better comparison test with my test.

 

[Booth.the.doberman]

Here are the tests I ran tonight:
#1. Go directly into a race (no practice or qualifying) with only the native cars. No fuel problems. Nothing unusual.
#2. Go directly into a race (no P/Q) with my cars. No fuel problems. Good race.
#3. Run a RaceEvent and do practice/qualifying/race with my guys. 14 out of fuel on first lap.
#4. Run a RaceEvent P/Q/R with just the native cars. Only six cars finished. 10 out of fuel on first lap. Six didn't run qualifying.
#5 Run a standard race with P/Q/R and just native cars. 9 finished the race. 7 dropped out on lap 1 with fuel. Only 11 of the 21 bothered to run qualifying.

I'm beginning to wonder about a parcferme thing. I've never touched the player.JSON file because I didn't think I ever needed to, so please don't tell me I must have changed something in there. I did not. It's as virgin as the day it was grabbed from Steam. When I searched on parcferme in that file it lists it everywhere as :3. That's supposed to be the RF2 default, it says, but I can't find anything that says what "default" is. I have no idea what that means.

By the way, in all the races above, I ran them without me on the track.

Also, it is consistent that when the safety car is out, the laps are not counted. If you're scheduled to run a 20-lap race and the safety car comes out and stays out for three laps, your final race will be 23 laps. It even says 23 laps at the end.

 

[Ace King]

Booth.....i see you're running a practice session before qualify and the difference is i only ran a 15 minute qualify before race time! Perhaps this difference is the smoking gun!

 

[Booth.the.doberman]

I also run 12 lap max qualifying since (if I remember correctly) that was the norm in the 90s...or you could run what you wanted to during the qualifying window. If parcferme is turned on, that could explain the sucking of the fuel. It doesn't explain the cars not going out to qualify or the stuff with the SC lap counting.

Would you mind testing a 30 minute practice, 30 minute qualifying (12 lap max) and 20 lap race? If you accelerate, it's not too bad. And I'll test the Q/R without the P on my end.

 

[Ace King]

I'll do your suggested test and report back.

ParcFerme description:

ParcFerme#":"0=off, 1=no setup changes allowed between qual and race except for 'Free Settings'), 2=same unless rain, 3=use RFM default 'ParcFerme=<0-2>'"

 

[Gijs van Elderen]

A RaceEvent or Vmod makes use of a .rfm file. The .rfm contains race rules for a specific series. If you choose the all cars and tracks rF2 makes use of the default .rfm. If you make a RaceEvent you can choose a custom .rfm or use the default.

I can only think that the laps with the SC on the track were not counted against the 20 laps.

Make sure count cautions laps = 0 in the .rfm
CountCautionLaps = 0 // Whether caution laps are counted towards the total number of laps; default=0, possible values 0 or 1.

ISI provides some .rfm file for a specific series. Maybe you can use the F3.5 .rfm to start with. http://imagespaceinc.com/downloads/rf2/rfm/FR352014SeriesrFm.zip

 

[Booth.the.doberman]

I tried eliminating the practice. The first test was my cars and 11 of them dropped out on lap 1 with fuel problems. So then I just put in the native ASR cars and tried again with Q/R only. This time 9 dropped out with fuel on lap 1. So I reduced qualifying to 5 laps and ran again with the native cars. This time 14 (FOURTEEN!) of them dropped out on the first lap and the race didn't finish because all the cars retired by lap 9.

I took the same setup to Sao Paulo. No problem at all.

 

[Booth.the.doberman]

I finally found the rfm file. For a RaceEvent, it was in a mas file inside the /installed/rfm folder. Once I extracted it, I saw that CountCautionLaps=0.

 

[Booth.the.doberman]

I'm dumbfounded. I'm weary. And I don't want to do this anymore. I don't want to have to edit parameters in different files all over the place to get a working adventure. It's like a Rube Goldberg machine. Jeez, I work with Solaris all day every day as an admin so I have an excellent grasp on user-unfriendly.

 

[Booth.the.doberman]

One more test. I deleted the ASR cars and downloaded them anew from ASR. Installed them. Absolutely no custom cars...strictly the ASR stuff. No special setups. 5 lap qualifying in 30 minutes, 20 lap race, 21 ai. RF2 checked on Steam for file validity.

Pos Up Driver Laps Time/Retired
1 ▲1 Riccardo Patrese 20 34:06.24
2 ▲18 Nelson Piquet 19 +1 lap
3 ▲10 Gerhard Berger 6 Accident
4 ▲15 Michael Schumacher 2 Accident
5 ▲5 Chuck Yin 2 Fuel
6 ▲5 Gunther Wendelton 2 Fuel
7 ▼6 Ayrton Senna 2 Accident
8 ▲1 Dolph Pete 1 Fuel
9 ▼2 Frank Fauvel 1 Fuel
10 ▼7 Alain Prost 1 Fuel
11 ▲3 Sal Babineaux 1 Fuel
12 ▲4 Gunnar Hops 1 DNF
13 ▼1 Sal Shabadoo 1 Fuel
14 ▼10 Gjerry Cornell 1 Fuel
15 ▲6 Edward Privett 1 Accident
16 ▼11 Devin Fuller 1 Fuel
17 ▼9 Max Angelo 1 Accident
18 ►0 Nigel Mansell 0 Accident
19 ▲3 Edward Dunn 0 Accident
20 ▼14 Dave Buskirk 0 Fuel
21 ▼6 Jean Alesi 0 Fuel

 

[Ace King]

Hey Booth.....i have the fuel issue resolved for this track. The track's AIW is at fault. I made many AIW's over the years in rF1 and during the final AIW process you need to allow the program to perform a proper "Fuel Usage" estimation. After i checked to see if the track's fuel usage was correct i discovered it was way off the mark. I'll need to properly package it all up and it's use will only be good for offline as it will cause a mismatch if anyone doesn't have this fixed AIW.

Note: If you share this fix with your friends then you could all race together online!

In my earlier tests the reason this fuel issue didn't expose itself was because i skipped both practice and qualify sessions and only tested the race. They apparently have enough fuel if you only do a race session. I did test the ParcFerme with it off and the fuel issue still took place, so that's when i decided to check the AIW which led to my discovery.

Another observation is that this track's AIW path leading out of the pits is also faulty because the line leads too close to the barriers which cause several AI severe damages which affect their qualify.

Another point is that when the safety car comes out the AI will follow the AIW's pace line which is also not the best line for an open wheel mod and the results are not pretty as they achieve more damages during the pace laps thus more accidents than necessary.

PS. When i get some free time (very busy) in the future i intend to fix these issues if the author hasn't done so.

Note: A temporary fix would be to turn the fuel usage off in the ingame settings!

Also, set flags to "full without "disqualify" and it will prevent the AI from acquiring a DQ during their Qualify if they go past the number of laps you have set or you could simply set the qualify laps to unlimited with whatever length of time you set which will accomplish this mission too.

 

[Booth.the.doberman]

That's great news. I'll just turn the fuel usage off. I have the flags set as you suggested because of the problem with running too many laps in qualifying. That still leaves me with the SC laps not being included in the normal laps, but since I'm not concerned about fuel usage, that doesn't really matter any more. Thanks so much for taking the time and having the interest to look deeper into this.

 

[Ace King]

Your are very welcome!

 

[Booth.the.doberman]

A followup. Tonight I went back to Monaco. Qualified fourth. I was running sixth on lap 15 when I got hit from behind. Final place 16. But no one ran out of fuel and there was only one accident and no safety car. So I think it's better now. We're off to Montreal next. It's raining there. I expect a **** finish there, too.

 

[Ace King]

You stated you were going to turn the fuel usage off as i suggested so perhaps is the reason why they didn't run out.

 

[hexagramme]

Good find, Ace. It sounded like an AIW issue to me too, so good that it's confirmed now.

 

[Ace King]

Option 2: Tested & Confirmed the AI won't run out of fuel except occasionally one or two.
Use "Unlimited" laps in your "Qualify" which gives the same results as Fuel Usage=Off