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

Hi Annick

> Is there some work left to update cars before to get them work perfectly
> with the modifications to SimuV2/V3 since 1.3.0 ?

Yes. That's the reason why there are releases at all, because if you get a
random CVS snapshot not all work might be done;-)

> If so, do you know which car works on which track and which not ?

No, this is subject of tests, but the reason for the "explosions" are the roll
bar and suspension settings.

> PS: Don't know if it the same pb, but with many tracks and Mac Laren F1
> or Jaguar XJ 220, the game freezes just before the ready message ...

If you see a black screen with a speed of 2e9 or something, then it is most
likely the anti-rollbar setting.

Bye, Bernhard

-- 

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

Hi, all.

After a full update on the code, I experiment a strange problem,
with Simu V2 and Simu V3 :
with the Mac Laren F1 car, on many tracks, except (at least)
for the CG Track 2, the car is in a strange position on the starting grid :
nearly at 10 m high, front to the ground ... and of course it falls !
No way to get control on it, it keeps making strange movements up and down
and rolling ...

When I switch back to Simu V2 1.3.0, all works perfectly well.

I don't know if it is helps, but with Simu V3, I get this messages
on the console :

Warning: car viper-gts-r, driver LLiaw: lift coefficients (0.900000, 0.600000), 
generate a lift of 3.000000, while maximum theoretical value is 0.785448
Warning: car cg-nascar-rwd, driver Tanhoj: lift coefficients (0.800000, 
0.500000), generate a lift of 2.600000, while maximum theoretical value is 0.820301
Warning: car lotus-gt1, driver Tita: lift coefficients (0.800000, 0.600000), 
generate a lift of 2.800000, while maximum theoretical value is 0.739245
Warning: car mclaren-f1, driver berniw 9: lift coefficients (0.900000, 
0.600000), generate a lift of 3.000000, while maximum theoretical value is 0.850591
Warning: car mclaren-f1, driver berniw two 9: lift coefficients (0.900000, 
0.600000), generate a lift of 3.000000, while maximum theoretical value is 0.850591
Warning: car mclaren-f1, driver bt 5: lift coefficients (0.900000, 0.600000), 
generate a lift of 3.000000, while maximum theoretical value is 0.850591
Warning: car mclaren-f1, driver Damned 8: lift coefficients (0.900000, 
0.600000), generate a lift of 3.000000, while maximum theoretical value is 0.850591
Warning: car mclaren-f1, driver Inferno 1: lift coefficients (0.900000, 
0.600000), generate a lift of 3.000000, while maximum theoretical value is 0.850591
Warning: car mclaren-f1, driver Olethros 5: lift coefficients (0.900000, 
0.600000), generate a lift of 3.000000, while maximum theoretical value is 0.850591
Warning: car mclaren-f1, driver JP MacLaren F1: lift coefficients (0.900000, 
0.600000), generate a lift of 3.000000, while maximum theoretical value is 0.850591

Is there some work left to update cars before to get them work perfectly
with the modifications to SimuV2/V3 since 1.3.0 ?

If so, do you know which car works on which track and which not ?

Thanks in advance.

PS: Don't know if it the same pb, but with many tracks and Mac Laren F1
or Jaguar XJ 220, the game freezes just before the ready message ...

Jean-Philippe.

Hello all,

Just got started with Torcs, and looking to develop a robot on my windows box, XP with vc++ 6.

I followed the instructions in the FAQ (download the bt robot package, rename all the files and entry point, etc.) but I'm stuck on "Add the myrobot project to TORCS, open the project with vc++ and try to compile it."

Exactly which files should I be adding to the Torcs project?
Thanks.

qin_23

_________________________________________________________________

Hello, guys.

I discovered Torcs some few weeks ago, and feel very enthusiastic with it.
Really impressive job !

As a quite "experienced" developper in real-time/non real-time simulation
(but not at all in gaming simulation, nor in 3D vizualization),
I would like to contribute, as much as I can afford from my limited free time ...

Is some of you able/willing to sum-up for me the current main streams
of development in Torcs, in order for me to help if desired,
and not to work seprately on something already current ?

I personnaly have some ideas, like :
- multi-threading, to increase frame rate on multi-core processors,
   (what's up with SDL switch ? is it ready ?)
- better (;-) support for joysticks/pads buttons, 8-ways controls,
   in Torcs configuration screens and during racing
   (be able no to touch the keyboard)
- better support for "wheel-gear-pedals" input devices :
   * non automatic clutch
   * grid gear changing
- separate human player configuration data (name, controls, ...)
   from the choice of a special car : in other words, when configuring
   a race, choose separatly a human player and a car
- quick race : be able to save multiple race configurations
   (my particular concern is to be able to race against the same set
    of robots, but changing easily the car for everyone, as an example).

Thanks in advance.

Jean-Philippe.
France.

Firstly, I have a book on aerodynamics, but it has not been much help.

> 
> 1) I've got some mix-and-match in the units side. The code seems to be 
> working mostly with SI units. However, the xml files have milimetres, 
> imperial (springs) and some others. Is there any list of what isn't SI in 
> the code? I'm basically doubting ride_height.
> 
I think the code understands many units.  No idea where the translation 
is though.  Why not try substituting mm with m, or inches, and see what 
happens?  I think it works OK.

> 2) Drag due to yaw =/=0 seems not implemented. See line 87. How much this 
> affects drifting? Would it be worth to run a couple of tests adding the 
> lateral area of the car into account or would I be wasting my time?
> 
I have this in simuv3.

> 3) On line 87, the drag reduction due to slipstreaming (dragK) is squared. 
> Is there any easy explanation for that?
> 
You are misreading the code.  dragK has two things integrated: the drag 
coefficient and the speed, so this is why it is squared.

> 4) Christos explains in his site that ground effect goes like
>        F_i= C_i*v²*exp(-b*h)
> It struck me to see ride height fully averaged instead of "axis averaged". 
> Is this intended for ease-of-setup? (balance is thus fully determined by 
> front and rear Clift).

Basically yeah, in fact it'd be difficult to determine what happens 
exactly when the front part is lower than the rear part, as the systems 
are connected.  Averaging everything is easier and good enough.

> 5) The actual ground effect formula seems to have the average height at the 
> fourth power. This differs from what is stated here, 
> http://www.idiap.ch/~dimitrak/torcs/carsetup.html but, again, simuv2 is 
> supposed to provide four times too much downforce. Is this the cause?
> 
no, the formula exp(-bh^4) just tells you how much the downforce 
_reduces_ when the car is further away from the ground. (my page has 
left out the ^4 factor).  The cause is simply in the selection of 
downforce constants.  For example, if you look at the claims made by 
ferrari about their 360-modena car, they indicate a total aerodynamic 
downforce factor that is around 4 times less than the one in torcs for 
this car.  Look also at the next question.


> 6) Wing efficiency L/D = wing->Kz / wing->Kx is 4.0. Is this high? Low? 
> I've been googling a short time, and the closest I've seen is a L/D list of 
> airplane wings and birds in wikipedia, and 4 is on the low side.
> 

L/D also depends on the angle of attack, wing size and geometry. 
However a simple upper bound on L/D is given by energy conservation and 
it depends just on wing size and angle of attack, and I am using it 
experimentally

> Finally, I'm sure all of you have noticed how even small amount of damage 
> has a large impact in laptimes in fast ovals and e-track 4, while one can 
> lap within 1 second or so in e-track 1 and g-track 2 and the other slow 
> tracks with damage upwards of 2000. Would it be very silly to implement a 
> bodywork downforce penalty similar to the drag penalty? The most obvious 
> effect I see is that robots might be penalised by not accounting for 
> downforce correctly.
> 

Well, I played for a bit with a 'randomizer' that made the wing forces 
deviating from just centrally applied backwards and downwards forces. 
This was really interesting in some cases as it really messed up your 
driving.

> Thanks a lot for your answers and your patiente if you made it this far. 
> Regards,
> 
> Miguel
> 
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> Torcs-devel mailing list
> Tor...@li...
> https://lists.sourceforge.net/lists/listinfo/torcs-devel

Dear folks,

I've been browsing aero.cpp in simuv2 (CVS torcs), and I've seen a couple 
of things that intrigue me, so maybe you'll be able to provide some 
answers. Please keep in mind that my knowledge in aero is pretty limited. 
Also, book and article references would be appreciated too.

1) I've got some mix-and-match in the units side. The code seems to be 
working mostly with SI units. However, the xml files have milimetres, 
imperial (springs) and some others. Is there any list of what isn't SI in 
the code? I'm basically doubting ride_height.

2) Drag due to yaw =/=0 seems not implemented. See line 87. How much this 
affects drifting? Would it be worth to run a couple of tests adding the 
lateral area of the car into account or would I be wasting my time?

3) On line 87, the drag reduction due to slipstreaming (dragK) is squared. 
Is there any easy explanation for that?

4) Christos explains in his site that ground effect goes like
       F_i= C_i*v²*exp(-b*h)
It struck me to see ride height fully averaged instead of "axis averaged". 
Is this intended for ease-of-setup? (balance is thus fully determined by 
front and rear Clift).

5) The actual ground effect formula seems to have the average height at the 
fourth power. This differs from what is stated here, 
http://www.idiap.ch/~dimitrak/torcs/carsetup.html but, again, simuv2 is 
supposed to provide four times too much downforce. Is this the cause?

6) Wing efficiency L/D = wing->Kz / wing->Kx is 4.0. Is this high? Low? 
I've been googling a short time, and the closest I've seen is a L/D list of 
airplane wings and birds in wikipedia, and 4 is on the low side.

Finally, I'm sure all of you have noticed how even small amount of damage 
has a large impact in laptimes in fast ovals and e-track 4, while one can 
lap within 1 second or so in e-track 1 and g-track 2 and the other slow 
tracks with damage upwards of 2000. Would it be very silly to implement a 
bodywork downforce penalty similar to the drag penalty? The most obvious 
effect I see is that robots might be penalised by not accounting for 
downforce correctly.

Thanks a lot for your answers and your patiente if you made it this far. 
Regards,

Miguel

Hello,

I'm using torcs as part of a research project, to simulate an environment
where cars follow each other in a platoon.
In order to do this, we have to make some assumptions about how cars are
able to behave, namely what is the stopping distance in case of a full
brake.

According to some sites (e.g. http://www.csgnetwork.com/stopdistcalc.html),
the distance can be computed by the formula:

d = (v*v) / 2ug, where u is the coefficient of friction between the tires
and the road.

As far as I know, the coefficient is something like 0.75 for a typical road
and tires. According to this site:

http://www.roymech.co.uk/Useful_Tables/Tribology/co_of_frict.htm

It should be between 0,5-0,8. According to the first site, "For many newer
high performance tires with good tread, the coefficient of kinetic friction
on a dry road surface may approach 0.8 if the braking is not so prolonged as
to cause tire melting."

If you input a 0.8 coefficient in the formula, for a 22 m/s speed you get a
stopping distance of ~30.87 m (you can test it on the first site).

d = (v*v) / 2ug = (22*22) / 2 * 0.8 * 9.8 ~= 30.87 m.

In our simulation using torcs, the cars are braking much faster, with a
stopping distance of about 9.67 m for the same speed. Here are the complete
values:

m/s  distance
02   0.055786
03   0.163696
04   0.296387
05   0.486694
06   0.715820
07   0.991577
08   1.297729
09   1.672729
10   2.091309
11   2.507965
12   2.891869
13   3.508484
14   3.956104
15   4.603813
16   5.303360
17   5.959892
18   6.602043
19   7.316177
20   8.078354
21   8.800949
22   9.675278
23   10.018005
24   10.463196
25   11.150696
26   12.195404

This gives us a coefficient of friction of about 2.75, which (as far as I
can know) is not realistic.

So, my question are the following:

- Is torcs supposed to be realistic about stopping distances?
- Are the above stopping distances to be expected for a car of category
'Track-RWD-GrA'?
- Is it possible to change any parameter so that we obtain stopping
distances similar to realistic non-racing cars?

Thanks a lot for your support!

Best regards,
Luís Marques

Donnar wrote:
> 
> Would it be possible to show a message on the startup screen, i.e. 'Press F1 
> for more options' or something like that ?
> 

That is definitely something we should put there.  I'll put it in the 
TO-DO list.

I have to confess, that I'm reluctant to learn when it comes to computer 
games. The simple keyboard interface together with the realistic physics is a 
reason why I prefer TORCS to other racing games. I mostly use TORCS for 
a 'Quick Race' during my lunch time.

However, although I know TORCS for quite a while now, I just recently 
discovered the 'Keys Definition' screen (F1) and the phantastic options 
hidden behind it.

Would it be possible to show a message on the startup screen, i.e. 'Press F1 
for more options' or something like that ?

Thanks,
Donnar