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

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Brian Martin wrote:
|
| I determined that this  problem was caused by stale bot code on my
| system. (I hope i did not waste your time on this problem,  Christos.
| If i did, you  have  my  sincere  apologies)
|

It is OK, I have but little time these days to spend on torcs, and
whatever I have I spent on the sim.

Part of the levitation problem could be the raceengine code.
The other part is the vertical collision code I think.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFKSR26yBv6kELa61sRAqaAAJ4/PnXGisZ1x6Qg4HMZdSFodEpTagCfaHw7
hr2g/+MD62O/n944EtykOn8=
=4d8/
-----END PGP SIGNATURE-----

I determined that this  problem was caused by stale bot code on my 
system. (I hope i did not waste your time on this problem,  Christos.
If i did, you  have  my  sincere  apologies)

The problem with the levitating car at the start of the race is still
there, however.

I  have  been  doing  nothing  but  working  on  torcs  for  the  last year
and  a half. My changes are spread all over the place. I can't just 
replace my version of torcs, i have to carefully merge the changes
in 1.3.1 into my version of torcs.

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


|>
| The code i have presented is not a real fix, but a work-around for the
| problems i was having with the new simuv2. I have tested with this
| code. Race results seem to be identical and i can't detect any
| difference in the way the cars drive when i drive the cars. The only
| detectable difference between the original code and my "fixed" code is
| that the problem i was having with levitating cars is resolved. As I
| stated in an earlier post,  i  would like to have a real solution to
| this problem that doesn't degrade the accuracy of the simulation.


I think that just the spring.packers fix should help a bit already.

| I think it would be  useful  to  reproduce  the  original  problem.  Try
| throwing together a car with the xml file i provided in my first post.
| (I can send you the entire car if you want.) Now start a race with the
| car and hit pause right away. You should see the car positioned slightly
| above the track. Look at the of value of new_susp_x in the debugger. I'm
| betting you will see that the value is way off. Now turn
| pause off and wait for the race state to change to race. You should see
| the car drop down on the track. Look at the value of new_susp_x. Now the
| value should be correct.

OK, thanks for doing this. I think I know what the problem is now. We
just have to have a good initial value for wheel->susp.x.

Setting it to
~        wheel->susp.x = wheel->pos.z - Zroad;
initially will probably work.

However, setting it to this value constantly has a side-effect: it makes
the wheel instantaneously move to touch the road. In reality a wheel
would sometimes not touch the road in bumps.

I think a good solution would be to use the Zroad hack only in the
RM_RACE_PRESTART phase of the simulation, which is the phase where all
the cars are placed on the track.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFKR2BFyBv6kELa61sRAiH1AJ9BXSR3h/Brh8wsQbZV3tUCBJXY7gCfV0mH
3faNQdZen8saSZgoSEV+VWE=
=wr5S
-----END PGP SIGNATURE-----

Christos Dimitrakakis wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> |     // Wheel susp.x is not the wheel movement, look at SimSuspCheckIn,
> | it becomes there scaled with
> |     // susp->spring.bellcrank, so we invert this here.
> |     //tdble prexwheel = wheel->susp.x / wheel->susp.spring.bellcrank;
> |
> |     // unreliable value tdble new_susp_x= prexwheel - wheel->rel_vel *
> | SimDeltaTime;
> |
> |
>
>
> Why are you commenting this out? This is not an 'unreliable value'. This
> is used to let the wheel move properly - otherwise the wheel's mass is
> not taken into account.
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.6 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>
> iD8DBQFKRplwyBv6kELa61sRAmBZAJ9RgLqvIydQ3+jjBE/XI3LDRPyMdwCeIhdQ
> PB327Ftcd4lWJm9XpYrX8Wg=
> =clq5
> -----END PGP SIGNATURE-----
>
The code i have presented is not a real fix, but a work-around for the 
problems i was having with the new simuv2. I have tested with this
code. Race results seem to be identical and i can't detect any 
difference in the way the cars drive when i drive the cars. The only
detectable difference between the original code and my "fixed" code is 
that the problem i was having with levitating cars is resolved. As I 
stated in an earlier post,  i  would like to have a real solution to 
this problem that doesn't degrade the accuracy of the simulation.

I think it would be  useful  to  reproduce  the  original  problem.  Try
throwing together a car with the xml file i provided in my first post.
(I can send you the entire car if you want.) Now start a race with the 
car and hit pause right away. You should see the car positioned slightly 
above the track. Look at the of value of new_susp_x in the debugger. I'm 
betting you will see that the value is way off. Now turn
pause off and wait for the race state to change to race. You should see 
the car drop down on the track. Look at the value of new_susp_x. Now the 
value should be correct.

I would like to provide you with a real fix for this problem. Although i 
am a software engineer by profession, i have no experience with physics 
simulations , so i am clueless when i look at physics engine code.

If there is anything you can think of that i can do to help resolve the
problem, i will be happy to help.

Brian

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

|     // Wheel susp.x is not the wheel movement, look at SimSuspCheckIn,
| it becomes there scaled with
|     // susp->spring.bellcrank, so we invert this here.
|     //tdble prexwheel = wheel->susp.x / wheel->susp.spring.bellcrank;
|
|     // unreliable value tdble new_susp_x= prexwheel - wheel->rel_vel *
| SimDeltaTime;
|
|


Why are you commenting this out? This is not an 'unreliable value'. This
is used to let the wheel move properly - otherwise the wheel's mass is
not taken into account.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFKRplwyBv6kELa61sRAmBZAJ9RgLqvIydQ3+jjBE/XI3LDRPyMdwCeIhdQ
PB327Ftcd4lWJm9XpYrX8Wg=
=clq5
-----END PGP SIGNATURE-----

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Brian Martin wrote:
| I have done a lot of testing since my last post. Two cars will behave
| unnaturally when they become air-born. The cars will stand up on one
| or two wheels and slowly settle on the track. The cars will then drive
| normally. Again, this  problem occurs with  a combination  of low mass
| and high spring pressure.
|

OK, this makes sense. I will test later and include in the main code.

| This  code  change will solve the problem:
|
| In wheel.cpp:
|
| void SimWheelUpdateRide(tCar *car, int index)
| {
|     tWheel *wheel = &(car->wheel[index]);
|     tdble Zroad;
|
|     // compute suspension travel
|     RtTrackGlobal2Local(car->trkPos.seg, wheel->pos.x, wheel->pos.y,
| &(wheel->trkPos), TR_LPOS_SEGMENT);
|     wheel->zRoad = Zroad = RtTrackHeightL(&(wheel->trkPos));
|
|
|     // Wheel susp.x is not the wheel movement, look at SimSuspCheckIn,
| it becomes there scaled with
|     // susp->spring.bellcrank, so we invert this here.
|     //tdble prexwheel = wheel->susp.x / wheel->susp.spring.bellcrank;
|
|     // unreliable value tdble new_susp_x= prexwheel - wheel->rel_vel *
| SimDeltaTime;
|
|
|         tdble max_extend =  wheel->pos.z - Zroad;
|     wheel->rideHeight = max_extend;
|
| //////////////////////test///////////////////
|     tdble prex = wheel->susp.x;
| //////////////////////////////////////////////////
|
|
|     if (max_extend < wheel->susp.x) {
|         wheel->susp.x = max_extend;
|         wheel->rel_vel = 0.0f;
|     } else if (wheel->susp.x < wheel->susp.spring.packers) {
|         //////////////test (fixes elevated car at race start for most
| cars)///
|         wheel->susp.x = wheel->susp.spring.packers;
|         /////////////////////////////////////////////////////
|         wheel->rel_vel = 0.0f;
|     ///////////////test fixes all mispositioned car problems/////
|     }else{
|         //wheel->susp.x = new_susp_x;
|         wheel->susp.x = wheel->pos.z - Zroad;
|     }
|     //////////////////////////////////////////////////////////////
|
| //    tdble prex = wheel->susp.x;
|
|
| //    wheel->susp.x = new_susp_x;
|
|
|     // verify the suspension travel, beware, wheel->susp.x will be
| scaled by SimSuspCheckIn
|     SimSuspCheckIn(&(wheel->susp));
|     wheel->susp.v = (prex - wheel->susp.x) / SimDeltaTime;
|
|     // update wheel brake
|     SimBrakeUpdate(car, wheel, &(wheel->brake));
|
| }
|
| This code will also solve the problem of simuv going berzerk with huge
| damper rebound values.
|
|
- 
------------------------------------------------------------------------------
| _______________________________________________
| Torcs-users mailing list
| Tor...@li...
| https://lists.sourceforge.net/lists/listinfo/torcs-users

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFKRpgTyBv6kELa61sRAnXgAJ9nmjpn/UsERIGHoBNtybIC93qLvwCgjx/6
XeootWSbW9MVf82ck9fsunw=
=bOQz
-----END PGP SIGNATURE-----

I have done a lot of testing since my last post. Two cars will behave
unnaturally when they become air-born. The cars will stand up on one
or two wheels and slowly settle on the track. The cars will then drive
normally. Again, this  problem occurs with  a combination  of low mass
and high spring pressure.

This  code  change will solve the problem:

In wheel.cpp:

void SimWheelUpdateRide(tCar *car, int index)
{
    tWheel *wheel = &(car->wheel[index]);
    tdble Zroad;

    // compute suspension travel
    RtTrackGlobal2Local(car->trkPos.seg, wheel->pos.x, wheel->pos.y, 
&(wheel->trkPos), TR_LPOS_SEGMENT);
    wheel->zRoad = Zroad = RtTrackHeightL(&(wheel->trkPos));


    // Wheel susp.x is not the wheel movement, look at SimSuspCheckIn, 
it becomes there scaled with
    // susp->spring.bellcrank, so we invert this here.
    //tdble prexwheel = wheel->susp.x / wheel->susp.spring.bellcrank;

    // unreliable value tdble new_susp_x= prexwheel - wheel->rel_vel * 
SimDeltaTime;


        tdble max_extend =  wheel->pos.z - Zroad;
    wheel->rideHeight = max_extend;

//////////////////////test///////////////////
    tdble prex = wheel->susp.x;
//////////////////////////////////////////////////


    if (max_extend < wheel->susp.x) {
        wheel->susp.x = max_extend;
        wheel->rel_vel = 0.0f;
    } else if (wheel->susp.x < wheel->susp.spring.packers) {
        //////////////test (fixes elevated car at race start for most 
cars)///
        wheel->susp.x = wheel->susp.spring.packers;
        /////////////////////////////////////////////////////
        wheel->rel_vel = 0.0f;
    ///////////////test fixes all mispositioned car problems/////       
    }else{
        //wheel->susp.x = new_susp_x;
        wheel->susp.x = wheel->pos.z - Zroad;
    }
    //////////////////////////////////////////////////////////////
 
//    tdble prex = wheel->susp.x;


//    wheel->susp.x = new_susp_x;


    // verify the suspension travel, beware, wheel->susp.x will be 
scaled by SimSuspCheckIn
    SimSuspCheckIn(&(wheel->susp));
    wheel->susp.v = (prex - wheel->susp.x) / SimDeltaTime;
   
    // update wheel brake
    SimBrakeUpdate(car, wheel, &(wheel->brake));

}

This code will also solve the problem of simuv going berzerk with huge 
damper rebound values.

Hi again Christos,

I solved the problem so simply:
Not enough suspension course, I increased all of the sides to 0.2 and the
car can change lanes on the speedways just fine.

Bye,
Haruna.

Ok, it's a kludge, not really a fix for some cars being
positioned wrong on the track. You can reproduce this problem with the xml
file i provided in my previous post. The problem is in wheel.cpp in 
SimWheelUpdateRide().
The problem manifests itself when

new_susp_x=prexwheel - wheel->rel_vel * SimDeltaTime;

generates a wrong value for new_susp_x. I would like to have a real 
solution to this
problem. The code changes are just a work around for the problem. I am 
not familiar
enough with the torcs code to even know where to begin looking for the 
root of this problem.

Here is the code:

void SimWheelUpdateRide(tCar *car, int index)
{
    tWheel *wheel = &(car->wheel[index]);
    tdble Zroad;

    // compute suspension travel
    RtTrackGlobal2Local(car->trkPos.seg, wheel->pos.x, wheel->pos.y, 
&(wheel->trkPos), TR_LPOS_SEGMENT);
    wheel->zRoad = Zroad = RtTrackHeightL(&(wheel->trkPos));


    // Wheel susp.x is not the wheel movement, look at SimSuspCheckIn, 
it becomes there scaled with
    // susp->spring.bellcrank, so we invert this here.
    tdble prexwheel = wheel->susp.x / wheel->susp.spring.bellcrank;

    tdble new_susp_x= prexwheel - wheel->rel_vel * SimDeltaTime;


        tdble max_extend =  wheel->pos.z - Zroad;
    wheel->rideHeight = max_extend;

//////////////////////test///////////////////
    tdble prex = wheel->susp.x;
//////////////////////////////////////////////////


    if (max_extend < wheel->susp.x) {
        wheel->susp.x = max_extend;
        wheel->rel_vel = 0.0f;
    } else if (wheel->susp.x < wheel->susp.spring.packers) {
        //////////////test (fixes elevated car at race start for most 
cars)///
        //wheel->susp.x = wheel->susp.spring.packers;
        /////////////////////////////////////////////////////
        wheel->rel_vel = 0.0f;
    ///////////////test fixes all mispositioned car problems/////       
        wheel->susp.x = wheel->pos.z - Zroad;
    }else if (wheel->rel_vel != 0.0f){
        wheel->susp.x = new_susp_x;
    }else{
        wheel->susp.x = wheel->pos.z - Zroad;
    }
    //////////////////////////////////////////////////////////////
 
//    tdble prex = wheel->susp.x;


//    wheel->susp.x = new_susp_x;


    // verify the suspension travel, beware, wheel->susp.x will be 
scaled by SimSuspCheckIn
    SimSuspCheckIn(&(wheel->susp));
    wheel->susp.v = (prex - wheel->susp.x) / SimDeltaTime;
   
    // update wheel brake
    SimBrakeUpdate(car, wheel, &(wheel->brake));

}

Thank you.

brian

Hello Brian,

I checked your setup and initally thought this was caused by too high damper
settings (rebound/bump).
When I checked the .xml, I think that such stiff suspension will cause the
car to basically act weird.
Maybe you could try setting the stiffness below 2000.
Also check your roll center heights and suspension courses, they should be
mostly the same as each other.

By the way, 1000+ nm in the lower ranges of the torque-band?
How the... do you handle that?

Bye,
Haruna.

i have an old version of torcs (2.4 from debian etch) , which has always 
worked.
(1.3.x does not work on my system). I updated simuv2 with version 1.3.1.
I'm very pleased with the new simuv, and it works fine with almost all 
my car
setups (i have 53)

However, with several setups the car will levitate on the track at the 
start, then drop down on the track and drive normally on go, or worse, 
rise up
and spin around on one wheel. The problem seems related to the 
combination of a very low mass car and very high spring pressure/bellcrank.

Am i an idiot for  using  simuv2 1.3.1 in 2.4 or is there a bug in simuv?

i have attached the xml file for one of the problem cars.

Thanks to all the torcs contributors.

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Haruna Say wrote:
| Hi all,
|
| While trying to make the cg-nascar-rwd drive more realistic, I noticed
| that if you turn right even slightly on the speedway tracks, you
| instantly lose control, and the car goes into a spin, and kind of
bouncing.
| It happens with all cars.
|

Is it simuv2? Then this is due to the physics engine not handling banked
tracks very well.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFKNysKyBv6kELa61sRAiq7AJ9ybGh2sMf7t7iJouI3AFZGIE312QCdFNSo
Jg6OEvj0PWN4/eR/IZrIRws=
=waUf
-----END PGP SIGNATURE-----

Hi all,

While trying to make the cg-nascar-rwd drive more realistic, I noticed that
if you turn right even slightly on the speedway tracks, you instantly lose
control, and the car goes into a spin, and kind of bouncing.
It happens with all cars.

Bye,
Haruna.

Hi Peter

> I have a track I've been working on dunno how well it might be received, it's one I made up to learn how to build them.
> It's kinda based on Vicente Martí Centelles's excellent tutorial ( Thanks Vicente - I'd have to say it took me a bit longer than 20 minutes though!)
> 
> If there was interest where would people recommend I host the track? 
> 
> I was hoping to post it to the TORCS Xchange but I see that there's a warning flagging upload errors. 
> 
> So what's the deal with new tracks for public consumption these days can anyone reccomend (free) file hosting sites? Or should I wait for the TORCS Xchange site?

There are several possibilities, I think TORCS Xchange is the way to go. If you
like I can host the track till it is up again (if it is released under the Free
Art License and is "legally clean", meaning no real advertisments, no textures
borrowed from dubious sources like commercil games, etc.).

You can host it as well yourself and I link to it from the TORCS page.

Let me know what do you prefer to do.

Best regards

Bernhard

Hi All,

My First Post! :)

I have a track I've been working on dunno how well it might be received, it's one I made up to learn how to build them.
It's kinda based on Vicente Martí Centelles's excellent tutorial ( Thanks Vicente - I'd have to say it took me a bit longer than 20 minutes though!)

If there was interest where would people recommend I host the track? 

I was hoping to post it to the TORCS Xchange but I see that there's a warning flagging upload errors. 

So what's the deal with new tracks for public consumption these days can anyone reccomend (free) file hosting sites? Or should I wait for the TORCS Xchange site?


Regards,

Sprintae86.



   




________________________________
From: Bernhard Wymann <be...@bl...>
To: Tor...@li...
Sent: Monday, 1 June, 2009 9:11:44 PM
Subject: [Torcs-users] Release 1.3.2, tracks

Hi all

If somebody wishes to submit some nice track for 1.3.2, then the time is now,
let me know. Thank you.

Bernhard

------------------------------------------------------------------------------
Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT 
is a gathering of tech-side developers & brand creativity professionals. Meet
the minds behind Google Creative Lab, Visual Complexity, Processing, & 
iPhoneDevCamp as they present alongside digital heavyweights like Barbarian 
Group, R/GA, & Big Spaceship. http://p.sf.net/sfu/creativitycat-com 
_______________________________________________
Torcs-users mailing list
Tor...@li...
https://lists..sourceforge.net/lists/listinfo/torcs-users



      Need a Holiday? Win a $10,000 Holiday of your choice. Enter now.http://us.lrd.yahoo.com/_ylc=X3oDMTJxN2x2ZmNpBF9zAzIwMjM2MTY2MTMEdG1fZG1lY2gDVGV4dCBMaW5rBHRtX2xuawNVMTEwMzk3NwR0bV9uZXQDWWFob28hBHRtX3BvcwN0YWdsaW5lBHRtX3BwdHkDYXVueg--/SIG=14600t3ni/**http%3A//au.rd.yahoo.com/mail/tagline/creativeholidays/*http%3A//au.docs.yahoo.com/homepageset/%3Fp1=other%26p2=au%26p3=mailtagline

Hi all

If somebody wishes to submit some nice track for 1.3.2, then the time is now,
let me know. Thank you.

Bernhard