rec.autos.simulators

A new Racer (free carsim) version is underway; I'm uploading it to
High Gear, where it will hopefully be up somewhere during tomorrow.

NOTE!! It is Windows only; I screwed up my Linux OpenGL environment a
bit, so I couldn't compile the Linux versions yet. Should be fixed
during the week. And because my main development machine is not in the
office, I cannot update the source code yet.

Racer v0.4.7
------------

v0.4.6.1 release notes (v0.4.6.1 never had its proper
release text):
- Fixes major grip bug; now it finally feels a bit realistic.
  You now get twice as much grip, like it should be.

v0.4.7:

NOTE:
- Windows-only version still because of missing development machines;
  Linux/SGI version will follow later.

Bugfixes:
- Modeler's ASE loader improved (had some bugs with some models,
  esp. with UVW unwrapping); also improved the optimizer to
  handle the new methods better.
- TrackEd & Modeler popup dialogs are now not overdrawn anymore.
  This problem was most apparent on Win98. It is a driver problem
  still, but I'll just circumvent it then. :(
- Debug output to COM1 is now optional (setenv QDBG_COM=1).
  This may resolve problem with wheels that are hooked on COM1.
- On Win98, a lot of people had trouble running Racer more than twice.
  My sound code is better now, in that COM is uninitialized at
  exit time. This is likely to fix the two-time running problem.
- FF jittering at low speed is damped more.

New features:
- Added a new friction combination method, courtesy from
  Gregor Veble for the idea, which gives better behavior,
  meaning more warning before the tire starts slipping.
  This friction method is now the default. It resembles
  very much the method from chapters 24/25 from Brian Beckman.
- Mipmapping, trilinear and anisotropic filtering support.
  See gfx.ini's new options.
- Aero drag for the body
- Wings aerodynamics
- Engine torque reaction on the body (like in GPL)
- Improved skid sound code (I think).
- Support for more than 1 controller
- Dual controller controlset for Logitech FF wheel
  and CH Pro Pedals (courtesy from Alison Hine)
- TSW2mod controlset (courtesy from Uwe Schuerkamp)
- Non-linear control support (variable from linear to non-linear)
- Improved the JoystImm utility (Win only) so it now handles
  more than 1 controller (you can select it).
- Improved TrackEd movement and selection
- Car selection screen
- Track selection screen
- Best laps are recorded per track (10 per track) and per driver
  (just 1 driver for now in data/drivers/default)
- Some transparency support for the cars. Still rudimentary,
  but working a bit (see the first screenshot on the
  website's news page).

Get it at http://www.racesimcentral.net/

Enjoy!
Ruud van Gaal

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.racesimcentral.net/
Free car sim  : http://www.racesimcentral.net/

Woh, you have put so much more stuff in there, awesome going!!  I will give
it a go when I get home, but now I feel that I should really do some work
when I get home! :)

All the best,
Ash

| Free car sim  : http://www.marketgraph.nl/gallery/racer/

About the new tyre method, I might have come to it independently, but
since Brian did it far before me (and it's partially in a book by Genta
as well from what I heard), I can't take any credit for it. I only might
have formulated it slightly nicer, without all the special sign
requirements, that's all the claim I can put in. Thanks for mentioning
my name anyway :).

-Gregor

Ah well, I did read about Brian's method, but it was quite a read and
sounded a little obscure. Your more direct formula approach made me
see the light. :)
The Genta book does describe some normalizing method, but it is
explained as somewhat experimental and that more work is done on the
subject. So that didn't give a really good feel, and there are a LOT
of divisions by potentially zero in there. Which is a shame.

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.marketgraph.nl/gallery/
Free car sim  : http://www.marketgraph.nl/gallery/racer/

Come on you guys, you are ***y killing me, could either Gregor or
Ruud, kindly explain this method in nice easy to swallow chunks. I
have just waded through Brians explaination, what WAS he smoking when
he wrote that??!?!? I must say that you physicist blokes seem to be
getting the best dope around :-)

I have no desire to just use the method as is from Racer as that
defeats the point of learning all of this stuff (although I am taking
a quick peak now to see how the method correlates to Brian's
explaination).

As we say here, Yoroshiku Onegaishimasu :-)

Colin

BTW Ruud, you rock man, all the best, you are like this one man coding
warrior, are you married? No married man (who wants to stay married)
has the time to do the amount of research / work that you are.

OK, if you ask this politely ;).

Basically, the idea is the following; there is no actual independent
slip ratio or slip angle acting on its own, but this are components of
the joint slip vector, and its the (appropriately normalized) magnitude
of this vector which determines the total force.

However, experimental data usually only gives you tyre behaviour in
ether lateral or longitudinal direction. Let's denote the lateral force
in uncombined operation F_y(S_y), where S_y is the slip angle, or
tan(slip angle), or sin(slip angle) or whatever you use (I use sin),
it's not really that important at smaller angles, and F_x(S_x) the
longitudinal force in pure longitudinal operation, with S_x being the
slip ratio. If you have a working sim, you probably have these functions
defined somewhere already.

How to combine the two? Let's introduce the total size of slip,

sigma = sqrt( (S_x/S0_x)^2 + (S_y/S0_y)^2 ),

where S0_x and S0_y are the peak values for slip angle/ratio (where the
force in uncombined operation is maximal). The symbol ^2 means the value
squared, of course. These can be rather pesky to determine from e.g.
Pacejka, but a rough estimate within about 20% error is most likely good
enough, the equations hardly critically depend on these parameters (when
in uncombined operation there is no dependence on them at all).

We'll now use the independent functions F_x and F_y to constuct actual
combined forces Fc_x and Fc_y from them. I'll just give the result here,
and explain it afterwards.

For the x direction, the result is (get your pencils out :) )

Fc_x = abs( S_x / (sigma * S0_x) ) * F_x( sigma * S0_x * sgn(S_x) )

and in the y direction, similarly

Fc_y = abs( S_y / (sigma * S0_y) ) * F_y( sigma * S0_y * sgn(S_y) ).

It's hard to explain the thought process on how to get to these
equations (and I'm not giving you the stuff we physicists are on :) ),
but let's look at why it's a good one.

Let's first assume that S_x=0, which means that the tyre is sliping
purely sideways (slip ratio=0). Fc_x is equal to 0 in this case because
of the abs () prefactor being 0. The slip size s equal to
sigma = sqrt( (S_y/S0_y)^2 )= abs( S_y/S0_y), and inserting this into
the equation for Fc_y will yield
exactly

Fc_y (uncombined) = F_y( S_y ),

which is just what we want. A simiar thing of course also holds for the
x direction.

On the other hand, in combined operation, and assuming for a second that
the peak force F_peak = F_x(S0_x) = F_y(S0_y) is (at least roughly) the
same for both the longitudinal and lateral direction, and the value of
sigma is 1 (this is the optimal slip) the size of the total force is,
after a few straightforward steps and taking the definition of sigma
into account to cancel the abs() prefactors,

F_total = sqrt( F_x ^2 + F_y ^2) = F_peak,

while looking at the abs() prefactors the result will show that it
points in the opposite direction to the total slip vector with
components (S_x,S_y).

It's also pretty straightforward to show that for small values of S_x,
S_y, the system behaves linearly and the same as if longitudinal and
lateral directions are taken separately, which they can be for very
small slips.

That's about it :).

Cheers,

-Gregor

Phew, glad that Gregor took this option. :)

Actually, it makes sense after a while; but it's a bit of magic that
it works so well. :) But the rules of overshooting your
slipratio/slipangle to cross the optimum peak are important here.

I'm not married. Do have a girlfriend though, and other girl-friends
as well. Certain things in life are more important the carsims. ;-)
Like the bottle of wine on the beach (with that girl) 2 days ago; you
can't beat that with the best powerslide, lol.

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.racesimcentral.net/
Free car sim  : http://www.racesimcentral.net/

|
| I'm not married. Do have a girlfriend though, and other girl-friends
| as well. Certain things in life are more important the carsims. ;-)
| Like the bottle of wine on the beach (with that girl) 2 days ago; you
| can't beat that with the best powerslide, lol.
|
|

Powerslide in what car? :)  lol

You are right of course, you just seem to do so much in so little time, I
need to get my brain accustomed to long periods of clear thinking and
concentration.  It is not easy to do.

Off to the movies tonight with my girlfriend - just to prove a point you'll
understand :)

Ash

Oh, well, maybe I'll do some further programming during the long
evenings then :).

-Gregor

--
Olav K. Malmin
remove .spam when replying

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.marketgraph.nl/gallery/
Free car sim  : http://www.marketgraph.nl/gallery/racer/

| On Wed, 01 Aug 2001 13:05:44 +0200, Gregor Veble
|
| ><sigh> all this talk of girlfriends, and I just had to take mine to the
| >train this morning as she went to Germany for a month...
| >
| >Oh, well, maybe I'll do some further programming during the long
| >evenings then :).
|
| We'll be expecting massive-multiplayer, AI and 50 new cars then after
| this month, Gregor! ;-)
|

Yeah gregor! Any Chance? :P

Ash

-Gregor

And if you skip dinner (no need to eat on your own), just implement
weather inbetween watching the news and the late night movie. :)

Ruud van Gaal, GPL Rank +53.25
Pencil art    : http://www.marketgraph.nl/gallery/
Free car sim  : http://www.marketgraph.nl/gallery/racer/