Drive to the limit, look at telemetry data of Alien Sim Racer and real Pro driver

Discussion in 'General Discussion' started by Joe, Dec 17, 2015.

  1. Joe

    Joe Registered

    Joined:
    Nov 14, 2013
    Messages:
    372
    Likes Received:
    21
    Thanks for posting. I am not a hardcore sim racer, unlike most of you guys. But this thesis is interesting to me, because he used Atze Kerkhof (alien sim racer) and Gustavo Menezes (pro driver, real car) telemetry data for comparison. See pg 21 of this thesis.
    (see interview video with Atze Kerkhof: https://www.youtube.com/watch?v=d2ur0L0LDOw)

    On Throttle: Kerkhof did almost “On-Off” way. I could not see a “soft easy” portion in the chart. They are very clean though. For the real pro driver, I do see a diff. As I learn, I was told that Throttle should NOT be an “ON-OFF” switch! But it is clear as shown Kerkhof did just like that.

    On Braking: the real driver’s data show more asymmetric than Kerkhof’s. That says the real driver used “trail braking”, while Kerkhof seems not (more or less “ON-OFF” switch). Could his pedal be too stiff?

    I like you guys thoughts on these. I may try myself on rF2 and post data to compare.
     
    Last edited by a moderator: Dec 17, 2015
  2. Led566

    Led566 Registered

    Joined:
    Jan 25, 2012
    Messages:
    358
    Likes Received:
    24
    Thank you for posting it.

    Don't want to hack your thread but this would be also a great place to discuss the "Spinelli steering lock problem", lately discussed in the Apex mod 3d.
    To make a long story short, Spinelli affirms that there is a flaw in the ISI physic engine that prevents to correctly simulate the steering behaviour of ANY car, i.e. in ANY ISI simulation to date, one only need a very reduced steering angle to turn compared to real life, and this especially in low speed turns.

    This affirmation is negated by some study that I posted in the other 3D, and now, I see, is falsified also in this very PhD thesis.

    If you scroll the thesis to page 76 - "Low Speed Turn simulation" you can read a complete description of what happens in reality compared to rF model.
    I just point your attention to figure 6.11 and table 6.5 - Steering sensitivity gradient
    The difference of gradient between Real Life and rF simulation is ...1.29%
    Steering angle seems perfectly comparable between real life and the rF mod.

    Now if someone has any real life MOTEC file to share, I would be very glad to make similar analysis with rF2, provided we could find a suitable rF2 mod.
     
  3. cosimo

    cosimo Registered

    Joined:
    Apr 14, 2013
    Messages:
    827
    Likes Received:
    99
    Noooooooooooooooooooooooooo...

    please no.

    not again :)
     
  4. hexagramme

    hexagramme Registered

    Joined:
    May 25, 2013
    Messages:
    4,245
    Likes Received:
    194
    Haha, it's about time his "theory" got properly challenged and not least buried once and for all, so I say bring it on. :p


    Sent from my iPhone using Tapatalk
     
  5. Guimengo

    Guimengo Guest

    The on/off portion is tied to the driver's personal style, but I feel it may be also related to the fact real life is dynamic, and a game has very defined parameters and very few fully dynamic situations. One aspect of aliens is that in a lot of games, especially older ones, they were kings of finding and exploring game exploits to their absolute limits and benefits.

    There has to be more coverage available but one old, interesting piece on telemetry was done with Schumacher and I think Herbert, at Benetton, and how they drove reasonably different on the same track. It's worth a search on the internet.
     
  6. Joe

    Joe Registered

    Joined:
    Nov 14, 2013
    Messages:
    372
    Likes Received:
    21
    Well, he did on rF1, I have not played rF1 for a while. I will assume he could have done on rF2 as well. I guess question would be: when starting to exit a corner, he could go full throttle right away. Is this (rF1 physics) realistic or not? The real F3 driver seemed could not do so, see the chart (Fig. 3.2).
     
    Last edited by a moderator: Dec 17, 2015
  7. Joe

    Joe Registered

    Joined:
    Nov 14, 2013
    Messages:
    372
    Likes Received:
    21
    I do not believe the "Updated rF model" is correct. It could be just over-fit the data by tuning the model parameters in this case (seemed "perfect" fit). If this model is true "correct", then it shall fit to other 3 cases as well, but it does not.

    There are two steering equations they used trying to compare between sim and real data:

    View attachment 18804

    with two cases: low speed cornering and high speed cornering. So they are talking about 4 cases (2x2), though.
    Their "Multibody Dynamic model" seems better to me. It fits 3 of them with 7%, -3%, and 10% error, respectively.
     
    Last edited by a moderator: Dec 17, 2015
  8. Led566

    Led566 Registered

    Joined:
    Jan 25, 2012
    Messages:
    358
    Likes Received:
    24
    I've read something similar also about a telemetry comparison between Schumacher and Barrichello.
     
  9. Minibull

    Minibull Member

    Joined:
    Sep 28, 2012
    Messages:
    1,556
    Likes Received:
    18
    Wasn't Senna one who would stab and blip the throttle on and off rapidly? Not quite a switch, but it goes against the ideas of smoothness for weight transfer, especially in F1
     
  10. arneschoonvliet

    arneschoonvliet Registered

    Joined:
    Apr 18, 2014
    Messages:
    50
    Likes Received:
    4
    tbh looking at the pic I don't see Atze getting on and off the throttle in an oscillating way. He is just driving bang on what the car is capable of doing. He is full throttle, gets off the throttle as quick as possible, brakes as hard as he can bleed off the brake paddle and then bang on back on the throttle.

    Where the real driver is braking earlier and as result of that less hard and bleeds the brake off a bit longer. Also he isn't bang back on the throttle, he needs to feed it in a bit in the real car otherwise he would get a to quick reaction of weight shift and with the tire not loaded up enough probably oversteer.
     
  11. Frenz

    Frenz Registered

    Joined:
    Aug 15, 2013
    Messages:
    27
    Likes Received:
    0
    There's a youtube you've probably seen of him driving an NSX at Suzuka and he definitely blips the throttle mid-corner in that.

     
  12. DurgeDriven

    DurgeDriven Banned

    Joined:
    Mar 20, 2012
    Messages:
    6,320
    Likes Received:
    43
    Who cares, are you going to fix it, is Spinelli.

    If you can't enjoy the developer engine as it is .............. give up or go play a better one !@!


    Don't you get tired of it.

    Barely anyones online you rather whinge rF2 is flawed.

    Why would anyone bother buying or onlining it the crap they read here.

    experts in your own minds tearing something down you have no hope in any world of ever doing yourself

    pathetic
     
  13. Minibull

    Minibull Member

    Joined:
    Sep 28, 2012
    Messages:
    1,556
    Likes Received:
    18
    Yeah that's one, and there's another one of the f1 cars. I think they were talking to his teammate, looking data, and there's a good trackside recording "braap, braap, braap" really quickly. The teammate says something like "I can't drive like that, it's all wrong, yet it works". Just goes to show how individual it is.

    In regards to that stuff above, I think Tim had been talking about how so many of us got used to pushing the front hard to go fast in the sims back in the days, and the sim not really punishing you for it. Whereas it seems the real racers can't, or don't, and focus more on the smooth entry and drive out, without scrubbing the fronts hard.
     
  14. Minibull

    Minibull Member

    Joined:
    Sep 28, 2012
    Messages:
    1,556
    Likes Received:
    18
    Found it, my mistake it wasn't a teammate, Palmer as a test driver.

     
  15. Joe

    Joe Registered

    Joined:
    Nov 14, 2013
    Messages:
    372
    Likes Received:
    21
    the x-axis in plot is distance instead of time. So we could not see the "lap time" in this case. But, it pin-points the both cars' positions action.
    If you look at carefully, the real pro driver leave no free foot. He started apply throttle as soon as off the trail braking much earlier, while Atze waits for (idle foot) and then full throttle ON. If there is lap time being recorded, I will think the real pro driver wins (at least for cornering time. Atze might be faster on straight line as shown). See:
    View attachment 18807
     
    Last edited by a moderator: Dec 18, 2015
  16. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    32
    Stop talking like that; "buried once and for all" as if you know for sure everything is perfect. You are the biggest, hardcore fanboy, just stop. Drop your pride and your ultra-loyalism. You've been banned from multiple sites and even your fake secondary accounts on those same sites were banned. Stop talking like such a fanboy.

    I can go around all of bathurst, Adelaide's final tight hairpin, Indy with the Nascars (not to mention with the absolute slowest steering lock upgrade) with literally less than 45 degrees of steering rotation. Go watch onboard laps of racing, it's not like that. The cars in game almost always suffer from "hyper-sensitive" dynamics this allows any car to be easily rotated and go around any corner without hardly having to turn the tyres.

    Even going at pathetically slow speeds, well under the car/tyre limit, you can easily make a car want to turn more and more sharply (which can eventually turn to oversteer).

    Even the F2000s I raced for a year, the Skippys, the Jim Russel F2000, and the Jim Rusell wingless F1600s I drove weren't that prone to oversteer. You would have 20 crashes a day at these schools even at ridiculously under-the-limit speeds (as most people drive them) if real cars behaved with such hyper-sensitivity.

    Cars in game bring a totally new meaning to steering with the pedals. The effect is insanely over-exaggerated and unlike any other sim ever (driven them all) and unlike anything in real-life.

    Part of the issue seems to be a phenomenon which is as though cars have no mass, no weight - no forward momentum therefore no sense of an object (vehicle) wanting to keep traveling in it's original direction of travel. This phenomenon means the cars can all-too-easily dart around and change direction of travel extremely instantly. The car wants to overly turn-in and travel in the direction the front-end points to rather than momentum wanting the car to continue on in the original direction of travel. That also leads to cars all-too-often wanting to sharply turn across the track laterally - more like an extremely sharp turn-in rather than "regular" oversteer - during oversteer situations and that is why so many oversteer moments look more like the car just instantly turns sharply and drives off across the track much more severely and often than any other sim as-well as real life

    I'm actually quite surprised a pretty good driver like you, Hex, is not able to perceive any of these things. Then again, perception and analysis isn't the same as driving according to what a particular physics engine likes/wants. Therefore, a pretty fast driver won't necessarily be able to perceive these sorts of details, and vice-versa, a slower driver isn't necessarily unable to perceive these sort of details
     
    Last edited by a moderator: Dec 20, 2015
  17. Lazza

    Lazza Registered

    Joined:
    Oct 5, 2010
    Messages:
    12,345
    Likes Received:
    6,572
    Could not or did not? If he couldn't, was it partly due to winter testing (less overall grip) ? How accurately can you judge the data, where the real life driver is applying very little throttle (let's say up to 30-40%) for 5-7 pixels on the graph before going to full throttle within 2-3 pixels, compared to the sim driver taking 4-6 pixels to reach full, or nearly full, throttle? How much of that difference (that can be judged) is due to driving style? How much of it might be due to different overall laptimes that you don't know? How accurate is the virtual track? Were there any local conditions on the real track that aren't accounted for in the sim version? Does the virtual F3 used do a good job of simulating realistic tyre trip and engine output? If the tyres don't act realistically, is that something that could potentially be fixed in rF or is it an inherent rF issue?

    There isn't even a steering trace, which could have helped explain the different throttle application (assuming similar laptimes).

    I hope you can see the point I'm making is there are a whole lot of variables at play here, and I wouldn't be jumping to too many conclusions either way. Other people would look at the same graphs and say hey, look how accurate rFactor is! I think you'd need a much more straightforward and repeatable test (say, a single flat fixed-radius turn with a given track width, in known good conditions) and be able to look at the full resolution telemetry data in order to start making any statements about accuracy.
     
  18. Minibull

    Minibull Member

    Joined:
    Sep 28, 2012
    Messages:
    1,556
    Likes Received:
    18
    @Spinelli: I'd have taken that post as just a little wee half hearted dig eh, not as a personal insult to your observations or you.

    I'd always be one to look for something actually showing it in the data somewhere. That's the only way this type of thing can be measured and figured out.

    At this level, you have offered your take on things without real hard proof, so therefore the responses, as far as I can feel, can be at exactly the same level.
    Hence, this never goes any further than each camp thinking they are right, until the next suitable (or unsuitable lol) thread pops up.

    Edit: As my observations, I've had the f1 car loose as a goose through the fast parts of Silverstone, and then with some setup changes, pushing and tight through the fast sections. Both had highly different lock applications. Loose car was sometimes near neutral or opposite lock.
    What I don't have is a trace of the actual wheel angles, which would be quite helpful in looking at this stuff.
     
    Last edited by a moderator: Dec 19, 2015
  19. matf1

    matf1 Registered

    Joined:
    Dec 27, 2011
    Messages:
    671
    Likes Received:
    2
    Man, when you say your thing, which I always try, really try to conceptualise, analyse... heck all the eyes!

    I remember this(not me):
     
  20. Nazirull Safry Paijo

    Nazirull Safry Paijo Registered

    Joined:
    May 28, 2011
    Messages:
    1,417
    Likes Received:
    35
    @Spinelli - can link me to a video or anything that can prove your "Steering Theory"?
     

Share This Page