torcs-devel Mailing List for TORCS - The Open Racing Car Simulator

Hi Christos,

That's great to hear about you !

Christos Dimitrakakis a écrit :
> Bernhard Wymann wrote:
>   
>> Hi all
>>
>> I reviewed SimCarUpdateSpeed and have the opinion that the rolling resistance
>> code in it must be removed, here is why (if no good arguments against this are
>> found or it causes terrible side effects I will do it soon):
>> - in SimCarUpdateForces the rolling resistance is already considered and
>> influences the acceleration
>> - in SimCarUpdateSpeed the new velocity is computed with the acceleration (which
>> already contains the rolling resistance), so we do it currently sort of twice
>> (it is a bit more complicated, the code has terrible side effects on surfaces
>> with large rolling resistance (try e.g. 0.12).
>>
>> So what do you think?
>>
>>     
>
> Checked simuv3, and it was indeed turned off nut in SimCarUpdateForces
> rather than SimCarUpdateSpeed - I will take a look at the differences
> and bring it in line with the simuv2.
>
> A suggestion.  Since most of the models are broken maybe it is now a
> good time to bring over the simuv3 differential patch to simuv2.
> I have a free day on Thursday which I could devote to looking at models
> and making sure they work.  Robots may take longer.  I'll look at
> Olethros first.
>
>   
Talking about Olethros robots, I did not succeded to have them working 
with VC++ 2005 Express.
I had to make some modifications to compile them, but they are crashing 
at the race start.
Can we check that problem Thursday too?

Bye,
Eric.

Bernhard Wymann wrote:
> Hi all
> 
> I reviewed SimCarUpdateSpeed and have the opinion that the rolling resistance
> code in it must be removed, here is why (if no good arguments against this are
> found or it causes terrible side effects I will do it soon):
> - in SimCarUpdateForces the rolling resistance is already considered and
> influences the acceleration
> - in SimCarUpdateSpeed the new velocity is computed with the acceleration (which
> already contains the rolling resistance), so we do it currently sort of twice
> (it is a bit more complicated, the code has terrible side effects on surfaces
> with large rolling resistance (try e.g. 0.12).
> 
> So what do you think?
> 

Checked simuv3, and it was indeed turned off nut in SimCarUpdateForces
rather than SimCarUpdateSpeed - I will take a look at the differences
and bring it in line with the simuv2.

A suggestion.  Since most of the models are broken maybe it is now a
good time to bring over the simuv3 differential patch to simuv2.
I have a free day on Thursday which I could devote to looking at models
and making sure they work.  Robots may take longer.  I'll look at
Olethros first.

Hi all

I committed the simuv2 changes into the CVS:
- Suspension improvement, car does not lift up any more magically on bumpy
  surfaces.
- Antirollbars are now functional.
- Rolling resistance is now correctly working (this fixes as well the
  "forever stuck in the sand" and the "wheel spinning like mad and
  car stuck" problems).

Beware, I did not adjust the categories and setups so far, so the simulation
might explode with several cars/robots. I will take care of this later.

Have fun,

Bernhard.

-- 

Visit my homepage http://www.berniw.org
Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb

> Well, without the code it's easier to get out of sand traps, this is the 
> only side effect I noticed. I vote for !

I'm happy with anything that helps you get out of sand traps :)  I know in
real-life that cars that go into the sand rarely get to come out again, but
in a game that's just not fun.  

> BTW, I played today with Andrews supercarsv2 set and the new simu, the 
> feeling is really great when driving !
> The sc-gt4 is my favorite, great sound !

Glad you like that set :)  I've got several more in the pipeline, however
they really need the revised championship feature that Mart Kelder & I am
working on to be finished first...

> I noticed that the sound of surrounding cars covers your own engine 
> sound when you play in internal view,
> that a little disturbing ;-) I'll have a look at this if no one objects.

I've always found that to be a bit odd as well.  It _does_ help make up for
the lack of peripheral vision (it'd be nice if TORCS had side mirrors
somehow)
as you can tell there's a car beside you.  

If you do mess around with the sounds, can I add some feature requests? 

1. When you're inside the car it'd be nice if it could look for an
"internal"
sound file.  That way you could have a sound that's damped down a bit when
you're in the enclosed cockpit compared to the louder exhaust noise you hear
outside.

2. Make it look for .wav files in the car folders first, then if it doesn't
find them there it can look in data/sounds.  It'd make it a lot easier for
people making 3rd party cars.

Btw I'm very close to having a rework of e-track-6 completed...

cheers
Andrew

Bernhard Wymann a écrit :
> Hi all
>
> I reviewed SimCarUpdateSpeed and have the opinion that the rolling resistance
> code in it must be removed, here is why (if no good arguments against this are
> found or it causes terrible side effects I will do it soon):
> - in SimCarUpdateForces the rolling resistance is already considered and
> influences the acceleration
> - in SimCarUpdateSpeed the new velocity is computed with the acceleration (which
> already contains the rolling resistance), so we do it currently sort of twice
> (it is a bit more complicated, the code has terrible side effects on surfaces
> with large rolling resistance (try e.g. 0.12).
>
> So what do you think?
>
> Thank you,
>
> Bernhard.
>
>   
Well, without the code it's easier to get out of sand traps, this is the 
only side effect I noticed. I vote for !
BTW, I played today with Andrews supercarsv2 set and the new simu, the 
feeling is really great when driving !
The sc-gt4 is my favorite, great sound !
I noticed that the sound of surrounding cars covers your own engine 
sound when you play in internal view,
that a little disturbing ;-) I'll have a look at this if no one objects.

Eric.

Hi all

I reviewed SimCarUpdateSpeed and have the opinion that the rolling resistance
code in it must be removed, here is why (if no good arguments against this are
found or it causes terrible side effects I will do it soon):
- in SimCarUpdateForces the rolling resistance is already considered and
influences the acceleration
- in SimCarUpdateSpeed the new velocity is computed with the acceleration (which
already contains the rolling resistance), so we do it currently sort of twice
(it is a bit more complicated, the code has terrible side effects on surfaces
with large rolling resistance (try e.g. 0.12).

So what do you think?

Thank you,

Bernhard.

-- 

Visit my homepage http://www.berniw.org
Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb

Hi Eric

> It seems that on certain conditions the cars can go (and stay) in weird
> positions (see the screen shots).
> I'll have a look at the code in case I can see something.

I have a simple reason for this: as example look at berniw two 2 (the p206),
there you can see the anti roll bar set to 10000, and the dampers are set very
low. Then keep in mind that the anti rollbar is a perfect spring (no damping).
So this is the expected behaviour (at least for the case I observed). In the old
version this does not matter, because the antirollbars did not work at all,
because a sign error in the set up.

So please check: Does it help to set the anti rollbar setting to 0 and the other
suspension values to reasonable settings (what is reasonable depends on the
combination of parameters, there is an expression which explodes in susp.cpp,
usually for fast 50-500 and slow 100-500 should work, depends of course on
bellcrank, etc. as well)?

Thank you for testing:-)

Bernhard

-- 

Visit my homepage http://www.berniw.org
Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb

Hi Eric

> Great ! I just tried it a little (not yet the anti-roll bar) and I 
> should say that the behaviour of the cars
> is much better now.
> I have a funny behaviour at the start, where it seems that the car don't 
> have enough time to land.

This happened to me as well in Quickrace mode, the problem there is that the
initial height is 1.0 m (the other race types have 0.2). Just adjust
quickrace.xml, this should do the trick (with "ordinary" suspension settings).

> I'll test more deeply the patch during the next days.

Great. I will try to fix as well the other issues (I think there is something
wrong with the rolling resistance, and the wheel spin goes mad under special
conditions (happens especially when the car gets stuck in the sand).

Bye, Bernhard.

-- 

Visit my homepage http://www.berniw.org
Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb

Great ! I just tried it a little (not yet the anti-roll bar) and I 
should say that the behaviour of the cars
is much better now.
I have a funny behaviour at the start, where it seems that the car don't 
have enough time to land.
I'll test more deeply the patch during the next days.

BTW, I have done some code cleanup (mostly type casts) to have TORCS 
compiling with VC++ 2005 Express.
I did not succeeded to compile correctly the olethros drivers, there is 
a crash at the start.
I'll submit the changes as soon as I can decently test them.

Thanks,
Eric.

Bernhard Wymann a écrit :
> Hi all
>
> This is a patch for the suspension code of simuv2, please test it.
> - Anti roll bars are now functional (just the "spring" parameter is used, the
> rest is ignored). With this you can adjust the cars over/understeering behaviour.
> - Added an improved version of Christos suspension fix (avoids the strange
> lifting on very bumpy sections, e.g. in the grass).
>
> Beware, the new code is now more sensitive to some parameters, e.g. if you run
> olethros 8 with the default setup the simulation explodes because of the insane
> suspension rebound values.
>
> To test the effect I attached an adopted definition for the p406. Feedback would
> be welcome.
>
> Bernhard.
>
>   
> ------------------------------------------------------------------------
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc.
> Still grepping through log files to find problems?  Stop.
> Now Search log events and configuration files using AJAX and a browser.
> Download your FREE copy of Splunk now >> http://get.splunk.com/
> ------------------------------------------------------------------------
>
> _______________________________________________
> Torcs-devel mailing list
> Tor...@li...
> https://lists.sourceforge.net/lists/listinfo/torcs-devel

Hi all

This is a patch for the suspension code of simuv2, please test it.
- Anti roll bars are now functional (just the "spring" parameter is used, the
rest is ignored). With this you can adjust the cars over/understeering behaviour.
- Added an improved version of Christos suspension fix (avoids the strange
lifting on very bumpy sections, e.g. in the grass).

Beware, the new code is now more sensitive to some parameters, e.g. if you run
olethros 8 with the default setup the simulation explodes because of the insane
suspension rebound values.

To test the effect I attached an adopted definition for the p406. Feedback would
be welcome.

Bernhard.

-- 

Visit my homepage http://www.berniw.org
Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb

Bernhard Wymann a écrit :
> Hi all
>
> I committed the 2 cars into the CVS, the car4 is just a re parametrized existing
> car, car5 is a totally reworked model.
>
> Bernhard.
>
>   
very good ! I tried them, they are nice to drive :-)

Eric.

Hi all

I committed the 2 cars into the CVS, the car4 is just a re parametrized existing
car, car5 is a totally reworked model.

Bernhard.

-- 

Visit my homepage http://www.berniw.org
Official TORCS racing: The TORCS Racing Board, http://www.berniw.org/trb

=D0=92 Fri, 5 Oct 2007 12:50:39 +0200
"Thorsten Heidt" <tho...@fa...> =D0=BF=D0=B8=D1=88=D0=B5=D1=82:

> I've downloaded stripe, compiled it and copied the files into the
> folder with the ac-files.
>=20
> After starting accc with this command:
>=20
> ./accc -g stadt4.acc -l0 stadt4-bak.ac -l1 stadt4-shade.ac -d3 1000
> -d2 500 -d1 300 -S 300 -es
>=20
> this error occured:
> sh: stripe: not found
> ./accc: line 31:  6178 Segmentation fault      (core dumped)
> $LIBDIR/accc-bin $*
>=20
> I also copied stripe to the bin- and the lib-folder - without
> success. :( (Stripe compiled without an error and I can run it with
> './stripe'. So I don't think it's a problem of stripe.)
>=20
> So - where is the problem? Does anybody know where I have to copy the
> compiled stripe-files? Do I have to add something to the
> accc-command, so accc can find stripe?

I don't know what is the exact behaviour of accc, but try and check if
stripe is your $PATH. You can check this by trying to launch it like
'stripe'. Sorry if this advise is too lame, but you only said that you
can launch it as './stripe'. :)

--=20
Alexander Tsamutali

Hi everyone!

I'm using Linux (Ubuntu 7.04) and TORCS 1.2.4 and I try to add shading to a
track.
I followed the instructions of the tutorial (thanks @ Vicente and all who
made this tut) and all went fine. I've got the original track file and the
'shading'-file.
Now I want to combine both in an acc-file.
I've downloaded stripe, compiled it and copied the files into the folder
with the ac-files.

After starting accc with this command:

./accc -g stadt4.acc -l0 stadt4-bak.ac -l1 stadt4-shade.ac -d3 1000 -d2 500
-d1 300 -S 300 -es

this error occured:
sh: stripe: not found
./accc: line 31:  6178 Segmentation fault      (core dumped)
$LIBDIR/accc-bin $*

I also copied stripe to the bin- and the lib-folder - without success. :(
(Stripe compiled without an error and I can run it with './stripe'. So I
don't think it's a problem of stripe.)

So - where is the problem? Does anybody know where I have to copy the
compiled stripe-files? Do I have to add something to the accc-command, so
accc can find stripe?

Thanks and greetz!
Thorsten

Bernhard Wymann a écrit :
> Hi José
>
>   
>> while playing with xrandr 1.2 , I noticed TORCS can't switch to 
>> fullscreen on
>> a randr 1.2 graphics driver (the ati 6.7.194 as example). It detects the
>> randr version and stops here.
>>     
>
> I doubt that this is a TORCS problem (but I may be wrong). Do you have Xinerama
> enabled? There is a similar problem with the NVidia drivers, but there the
> problem is in the graphics driver, not in TORCS.
>
> http://www.berniw.org/trb/forum/showthread.php?topicid=1088
>
> Bye, Bernhard.
>
>
>   

In fact, you are right : I used a Virtual 2048 768 line to enable randr 
1.2 separating or not the two 1024x768 screens.
But even if I switch to clone screen mode, fullscreen in 1024x768 
doesn't work. Interestingly, full screen at 800x600 works!

Anyway, this is with the free ati driver, the problem should be the same 
with intel one as all drivers will switch to randr 1.2.

José