AI pit stop strategy concerning tyre change

Discussion in 'General Discussion' started by Hygge, Aug 17, 2018.

  1. MRGreeny74

    MRGreeny74 New Member

    Joined:
    Aug 4, 2018
    Ratings:
    +1 / 0 / -0
    Come from a Race against the 100 % AI (WTCR) in Indianapolis (30 Lap Race, 4x Tire Wear, 4x Fuel Usage. no mandatory pit stops).......

    I had 2 pitstops and only changed the front tires each time i came to the pits. The AI (all 24 Opponents) had 3 pitstops and i won the race by a gap of 55 sec.

    The Pitstratey of the AI needs A LOT of work.
     
    • Informative Informative x 1
    • Useful Useful x 1
  2. Alex Hodgkinson

    Alex Hodgkinson KW Studios Developer

    Joined:
    Jun 5, 2017
    Ratings:
    +1,948 / 0 / -0
    This would imply that WTCR AI tyre wear is probably too high.

    Thank you, I'll work on it.
     
    • Like Like x 1
  3. Rujasu

    Rujasu New Member

    Joined:
    Jun 27, 2015
    Ratings:
    +4 / 0 / -0
    Since this is the thread that pops up when you search for Raceroom ai tyre strategy on the interwebs, is there any chance we could get a more definitive answer on how the AI determines tyre choices? Especially since the Group C cars will come with two sets of tyres, and I'd love to not run away from them in a longer race just because they do something stupid with pit stops.

    How does the AI choose which tyres to start with? Brief testing with the X-17 seems to indicate they take softs in a 10 minute race, mediums in a 20 minute race, and hards on 30 and up. Is that all the logic there is to it? In addition, in a 45 minute race with 2x wear, they drove hards, then pitted for another set of hards with 10-15 minutes left. Can it be assumed that they don't ever pick a different compound during a race?
    They do seem to at least pick the softest compound when wear is off, so that's something.
    Also from what I've seen, letting them practice seems to make no difference, at least with the X-17, so no data gets carried over from practice to race?

    Finally, a bit off-topic, but if the tyre strategy doesn't extend beyond the AC-style "here's a file that says how long each tyre can go, compare it to race length", is fuel any 'smarter'? How do they determine how much fuel to take? They at least seem to be less prone to strategic mistakes when only fuel use is on.
     
  4. Alex Hodgkinson

    Alex Hodgkinson KW Studios Developer

    Joined:
    Jun 5, 2017
    Ratings:
    +1,948 / 0 / -0
    I've noted your post and will give a proper response tomorrow morning.
    Give me a poke if I haven't done so by midday.
     
    • Like Like x 3
  5. Rujasu

    Rujasu New Member

    Joined:
    Jun 27, 2015
    Ratings:
    +4 / 0 / -0
    Consider yourself poked, good sir.
    In all seriousness, though, it's hardly an emergency, so a response is appreciated whenever you have the time.
     
  6. Alex Hodgkinson

    Alex Hodgkinson KW Studios Developer

    Joined:
    Jun 5, 2017
    Ratings:
    +1,948 / 0 / -0
    With regards to tyres you've correctly assumed that there is a pre-defined estimated life expectancy. Be it 15 minutes, 20 minutes or half an hour, the AI will choose a tyre which is guaranteed to last. When it comes to tyre changes, I believe that they will in certain situations swap compounds if the race time remaining suits that compound.

    When it comes fuel use, the AI use a fuel useage estimate which is exactly the same value used in the garage screen to give your lap value displayed next to litres. Accuracy here depends on how well that estimate has been set by yours truly or one of my predecessors.
     
  7. Rujasu

    Rujasu New Member

    Joined:
    Jun 27, 2015
    Ratings:
    +4 / 0 / -0
    Gotcha, thanks.
     
  8. Nico Kunze

    Nico Kunze Well-Known Member

    Joined:
    Sep 1, 2018
    Ratings:
    +395 / 0 / -0
    Since i got no answer on youtube ill try it here as well :D
    When you speak of a tire thatll last 20 minutes does that mean that after 20 minutes its on 0% or some other figure where youre just horribly off the pace?
     
  9. Alex Hodgkinson

    Alex Hodgkinson KW Studios Developer

    Joined:
    Jun 5, 2017
    Ratings:
    +1,948 / 0 / -0
    Firstly, the 20 minute value is very much a guide and worse case scenario. It could well be that when player driven, on the right kind of track it may last up to 25 minutes, or maybe the total opposite and only 15.

    A 20 minute tyre is going to be showing 0% wear after 20 minutes.
     
    • Informative Informative x 1
  10. Nico Kunze

    Nico Kunze Well-Known Member

    Joined:
    Sep 1, 2018
    Ratings:
    +395 / 0 / -0
    Thanks for clearing that up :)
    Now im wondering which tire/car wears the fastest in r3e to see how long id last on it :D
     
  11. rad

    rad Well-Known Member

    Joined:
    Jan 29, 2015
    Ratings:
    +458 / 0 / -0
    I'd guess FRX17 or Formula US on softest compounds
     
  12. Rujasu

    Rujasu New Member

    Joined:
    Jun 27, 2015
    Ratings:
    +4 / 0 / -0
    I did a 60 minute TCR race on Mid-Ohio with 3x (I think) tire wear, IIRC. The AI opted for a four-stop strategy, with the first pit stop occurring around when my own wear indicators started turning yellow. While it wasn't the worst strategy in the world, three stop was clearly faster.
    I see WTCR has had some fuel estimation tweaks since August, but I don't see anything about tires. How's everyone else's experience with them?
     
  13. Nico Kunze

    Nico Kunze Well-Known Member

    Joined:
    Sep 1, 2018
    Ratings:
    +395 / 0 / -0
    Gotta drive the us for a longer stint but the fx17 on soft tires doesnt drop below 60% on 15 minute runs for me so they should definitely last half an hour. If that indeed is the fastest wearing tire alex cjose some strange figures for his example :D
     
  14. ravey1981

    ravey1981 Well-Known Member Beta tester

    Joined:
    Apr 15, 2018
    Ratings:
    +873 / 0 / -0
    The fuel estimates in game are miles out on most if not all classes. Comparing the setup garage estimate to real time calcs done by crew chief or other apps shows this quite clearly. Results in the AI taking more pitstops than they actually need to and ruins longer races. The AI is not great on tyres either, endurance races have to be done with tyres and fuel off or there is just no point. Player has a massive advantage.
     
    • Agree Agree x 1
  15. Alex Hodgkinson

    Alex Hodgkinson KW Studios Developer

    Joined:
    Jun 5, 2017
    Ratings:
    +1,948 / 0 / -0
    Yep, there's definite room for improvement there and I have plenty of reports of poor AI fuel usage. Seeing as I'm at the helm of all of those numbers, it is useful to have that sort of thing brought to my attention.
     
    • Like Like x 2
  16. dungbeetle

    dungbeetle Well-Known Member

    Joined:
    Feb 15, 2016
    Ratings:
    +66 / 0 / -0
    Alex,
    Here's a thread I started back in December '18 which might be of help in solving the AI vs Player fuel consumption problem.

    https://forum.sector3studios.com/in...ngs-for-endurance-races-against-the-ai.12145/
     
    Last edited: Mar 28, 2019