1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Hillman Imp

Discussion in 'Racer Cars' started by boomer541, Sep 2, 2011.

  1. Made a few years ago and just updated, but there appears to be a problem perhaps with the car.ini file.

    When driving with ai on QCM's Arnaz track (copied default.ai and renamed imp.ai) the car oscillates and spins off the track. It runs ok on Carlswood and most other tracks.

    If you type "graph 0.1 -2...2 steering.pos" in the console you can watch the oscillation.

    Link to download:
    http://personal.my180.net/~hojaboom/links.html

    Download the ,gif file and rename it to .7z as my ISP doesn't like to send .7z files.
     
  2. Would like to check it out for sure, but I can't seem to get any of the files from your site downloaded. They either show up as 0kb files or throw errors when attempting to download right away, tried it in several different browsers. Perhaps you could upload it to one of the usual places like megaupload, speedyshare etc... ?
     
  3. KS95

    KS95
    RACER Moderator

  4. Thanks for that :)

    I can't claim to have found the solution to the spin-out behavior yet, though a first hunch would be that these are overshoots triggered by the slightly wavy nature of the AI lines plus the very direct responsiveness of the vehicle. Since the AI knows no leeway when it comes to following the path, unlike a human that looks ahead and realizes they can gently move back onto the racing line in a certain manner, the AI tries to remedy errors aggressively. Together with this particular vehicle's setup (CoG at ground level, sticky tyres and so on), that might just be too much at the moment. Lowering the ai.oversteer_correct/understeer_correct values can help a little bit.

    Basically all AI cars will show this kind of behavior to some degree, but it's certainly not the norm to have it reach these levels - more testing needed ;)
     
  5. I did try to alter the ai settings in the racer.ini file and tey helped just a little but not enuf. I tried to do an ai learn but that didn't work either.

    Prpblem only appears on certain tracks and not on others. carlswood works just fine, could be something for Ruud to look into if he has the time. Real bugger.

    The link is under the image of the imp. It is a .gif file that needs to be changed to a .7z file then opened with 7-Zip.
    The download then works with MSIE as I tested it.
     
  6. Thx Boomer, funny car sound + looking !
     
  7. The sound is from a lawn mower, lol. Looks like the real one I owned.
     
  8. I like drive this little car, boomer!
     
  9. Reading your latest post in the v0.8.35 release thread about the wheel lifting, I thought it might be helpful to give some more feedback. You mention that the "outside" front wheel is supposed to come off the ground during cornering, but the "trailing link" suspension currently makes it bounce instead of staying in a more steady condition. My guess would be that you meant the "inside" front wheel and the swing arm suspension of course, since the trailing links are in the back of the Imp :)

    To the point though, I think you'll continue to have trouble getting reasonable handling characteristics and finer points of the driving behavior such as the wheel lifting with the current centre of gravity vertical position. A proven rule of thumb is to take 37% of the vehicle's roof height as a good approximation for the static CoG height. With the 1.384m (or 1.374m ~= 54.1", depending on which number is closer) provided in the car.ini file, that would mean about 0.51m. The Imp currently has a CoG vertical position of 0.022m, or in other words, at ground level. As a result, there's almost no longitudinal load transfer and in the lateral direction, with roll centres being at -0.018m in the front and -0.034m in the rear, very little side to side transfer too. Leaving the numbers aside for a moment, that fact is also very obvious when driving the vehicle - it doesn't pitch or roll in a perceivable manner and thus won't be able to lift the front inside wheel either, nor does it give the driver any kind of feedback.
    Just by moving the CoG up into a range that seems sensible for the standard model - or even a bit lower for a race version like this - transforms the handling significantly and you can suddenly drive around in circles on three wheels for as long as there's fuel in the virtual tank :)

    Of course, changing the CoG height has a massive influence on the entire setup, so some other oddities that weren't visible before suddenly get highlighted. I'm sure you've known the http://www.imps4ever.info/tech/index.html website for a while already, they have some useful suggestions and data on the suspension as well. Roll centre height on the front being near the swingarm pivots as expected, as well as a much lower rear roll centre height for the trailing link rearend are listed as important characteristics for the balanced nature of the Imp's chassis setup.

    Having the basic elements such as the CoG position, masses and dimensions right before anything else is done to the setup seems vital to me - otherwise you end up chasing your own tail.



    I also had a glimpse at the qlog, you could check if your racer.ini has "note_ini_unused=1" set on top of "note_ini_fallback=1" (they're right next to each other at the end of the file) - there are some obsolete entries in car.ini at the moment.

    With or without correctly moving suspension, this will be a fun vehicle to have in Racer with a couple of tweaks. Starting with that CoG, other things will be much easier to see when the basics are in the ballpark and the real setup work can begin :)



    Finally, I had another go at downloading from your website with my mobile and it worked there - which made me try out if the security software on my PCs might be the cause... and guess what, indeed it recognized that the file format didn't match up with the contents (your renaming of .7z to .gif), blocked it but only gave a silent alarm somewhere in a logfile ^^
     
  10. Thanks for the info Cosmo I'll look at the stuff and try to get it better.