gmane.games.torcs.devel

Re: AI research - reinforcement learning

Daniele Loiacono — 2013-04-22T14:44:03

Dear Walter,

Bernhard already pointed you in the right direction.

I wuold just want to add few information.
In the literature, especially in the CIG (Computational Intelligence in
Games) research area, there are several works about applying machine
learning and evolutionary computation to TORCS.
Most of this works apply evolutionary computation techniques, but you can
find also some RL applications (e.g., our work at CEC 2010:
http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5586191&tag=1).

Concerning the software, we built an API on top of TORCS to run a
scientific competition. You could have a look to the page of this year
competition: scr.geccocompetitions.com and, maybe, have a look to the
papers about Simulated Car Racing competition published in the past years.

Best,
Daniele



On Mon, Apr 22, 2013 at 7:22 AM, Bernhard Wymann <berniw< at >bluewin.ch> wrote:

------------------------------------------------------------------------------
Precog is a next-generation analytics platform capable of advanc

Re: AI research - reinforcement learning

Bernhard Wymann — 2013-04-22T05:22:28

Hi Walter

Maybe you found it already, but I can give you some pointers:
- http://scr.geccocompetitions.com/previous-scr/ (I think there is one 
little "paper" available for each submitted opponent)
- Google/ Google Scholar 
http://scholar.google.com/scholar?q=torcs+car+reinforcement
- http://groups.google.com/forum/#!forum/racingcompetition
- You can search as well torcs-users/devel with sf.net 
http://sourceforge.net/p/torcs/mailman/?source=navbar

Not about learning, but about using TORCS:
http://www.youtube.com/playlist?list=PLq-L0hhLuRI23uh_HvNTnOD07esKsZkrt

Other experiences and thoughts?

Best regards

Bernhard


On 04/22/2013 05:37 AM, jlwalter wrote:


------------------------------------------------------------------------------
Precog is a next-generation analytics platform capable of advanced
analytics on semi-structured data. The platform includes APIs for building
apps and a phenomenal toolset for data science. Developers can use
our toolset for easy data analysis & visualization. Get a free a

AI research - reinforcement learning

jlwalter — 2013-04-22T03:37:49

Hello everyone,

My  name is Jeremiah and I'm  starting a research project on Reinforcement Learning testing different  algorithms' effectiveness at 'learning to drive', or dealing with  continuous actions. And I'm going to use TORCS as my simulation!

So  my question is, has anyone done any similar Reinforcement Learning research and have you any  tips or recommendations when it comes to setting the simulation  environment up?
I appreciate any help I can get!

Thankyou,
Jeremiah Walter
BComp, University of Tasmania

------------------------------------------------------------------------------
Precog is a next-generation analytics platform capable of advanced
analytics on semi-structured data. The platform includes APIs for building
apps and a phenomenal toolset for data science. Developers can use
our toolset for easy data analysis & visualization. Get a free account!
http://www2.precog.com/precogplatform/slashdotnewsletter_______________________________________________
Torcs-devel mailing list
Torcs-devel

[GECCO 2013] Simulated Car Racing Competition

Daniele Loiacono — 2013-04-17T20:48:53

Our apologies if you receive multiple copies of this announcement.
---------------------------------------------------------------------

               Simulated Car Racing Competition

                  to be held as part of the

  2013 Genetic and Evolutionary Computation Conference (GECCO-2013)
                July 6-10, Amsterdam, The Netherlands

                  Organized by ACM SIGEVO
               http://www.sigevo.org/gecco-2013

 Submission Deadline: June 21, 2013

---------------------------------------------------------------------

We are pleased to announce the 2013 Simulated Car Racing Competition.

Get ready to enter the challenge and to submit the fastest driver!


** GOAL **

The goal of the championship is to design a controller for a racing
car that will compete on a set of unknown tracks first alone (against
the clock) and then against other drivers. The controllers perceive
the racing environment through a number of sensors that describe the
relevant features of the car surroundings

brian gr

brian gr — 2013-02-27T10:17:14

http://www.decomagic.com.au/mx/iuqrm8t8hdt/s5tbaoid5udwhtq86qkwg.e&nbsp; 
------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_feb_______________________________________________
Torcs-devel mailing list
Torcs-devel< at >lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/torcs-devel

Re: Problem in running TORCS after building a robot

Bernhard Wymann — 2013-02-21T22:15:32

Hi Krzysztof

The robotgen script in 1.3.4 is outdated, this will be fixed for 1.3.5. 
Have a look at this post here:
http://sourceforge.net/mailarchive/forum.php?thread_name=50F17529.5040604%40bluewin.ch&forum_name=torcs-users

The top post contains a patch.

Or you can download the fixed robotgen version from here:
http://torcs.cvs.sourceforge.net/viewvc/torcs/torcs/torcs/robotgen?view=log&pathrev=r1-3-1

Btw. revision r1-3-1 is correct, I am still on this branch.

Hope this helps.

Best regards

Bernhard


On 02/21/2013 10:46 PM, Krzysztof W wrote:


------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_feb

Problem in running TORCS after building a robot

Krzysztof W — 2013-02-21T21:46:16

Hi everyone!

I would like to start developing a robot in TORCS, but I've got some
problems.

Few days ago I installed TORCS 1.3.4 from sources (all-in-one) on Ubuntu
12.10. The game runs perfectly. Then I built my own robot using "robotgen".
I hadn't any errors during a robot installation. But unfortunately, when I
start to play TORCS and I choose "configure race" to select my robot, I
receive an error and the program quits. I'm trying to make it work for a
long time, but I can't find any solution to this problem.

Below is a log after running "torcs -d".

Do you have any ideas where I've made a mistake?



GNU gdb (GDB) 7.5-ubuntu
Copyright (C) 2012 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-linux-gnu".
For bug reporting instructions, p

Re: 64 bit machine cpu time donation

Bernhard Wymann — 2013-02-08T15:41:53

Hi Megaf

Glad you like TORCS:-)

Do you mean the 1.3.0 binary installer? I did not do another one because 
there was practically no feedback and not many downloads, so I abandoned it.

I don not plan to do any distribution specific packaging, distribution 
people can do this better, and there are just too many distros out there 
(and many distributions package TORCS).

So thank you for your offer, but I have no use for the VM's (I run 64 
bit native, but for 32 bit Windows testing/development I use as well 
virtualized environments).

Best regards

Bernhard

On 02/07/2013 11:11 PM, Megaf wrote:


------------------------------------------------------------------------------
Free Next-Gen Firewall Hardware Offer
Buy your Sophos next-gen firewall before the end March 2013 
and get the hardware for free! Learn more.
http://p.sf.net/sfu/sophos-d2d-feb

64 bit machine cpu time donation

Megaf — 2013-02-07T22:11:09

Hi folks, For long I' ve been playing Torcs, I'm a big fan of it.
Now a days there are quite a lot of people with 64 bit CPUs and distros and
this number will get higher and higher.

Today I tried to get the binary linux build of Torcs to run here on my 64
bit Debian Squeeze PC, but I could not find a 64 bit version of it and the
version I found seemed to be broken.

Thinking about that I realized I could somehow help this fantastic project.
I am not a developer and my programming skills are just not good enough.

My PC is an AMD Phenom II x4 at 3,2 GHz with 8 GB of ram, there is a domain
that points to it already,
I don't use all it's power so I'd like to donate one or more virtual
machines to you so you could use it to build newer versions of torcs for
Linux, binary packages for 32 bit and for 64 bit.

For now, I have already up 3 VMs,
Each one got two cores and two GB of ram with 38 GB for / that you can use
to build and develop Torcs.
Debian Squeeze 64 bit, Debian Wheezy 64 bit and Fedora 18 64 bit. I'd

Re: System stability and changing gears in TORCS.

Bernhard Wymann — 2012-12-14T16:50:43

Hi James and all


I tries to fix this, could you try again? If the site behaves otherwise 
weird please let me know.

Best regards

Bernhard

------------------------------------------------------------------------------
LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial
Remotely access PCs and mobile devices and provide instant support
Improve your efficiency, and focus on delivering more value-add services
Discover what IT Professionals Know. Rescue delivers
http://p.sf.net/sfu/logmein_12329d2d

TORCS in VS2012, managed C#

SteveO — 2012-12-13T15:24:00

Hi All,

I hope this is the right place for this post - I am new to TORCS and just
starting a project in Visual Studio 2012 with managed C# that will be
accessing many of the TORCS physics functionality. I have TORCS compiled
and running in debug from VS2012 and can share my experiences if anyone
else is planning to do the same.

I have been looking at different ways to get access from managed C#.
Writing a wrapper is probably the right way but it would be very boring and
laborious. ( Unless of course *a bit cheeky* but does anyone have a wrapper
they are willing to share? )

Has anyone else tried a different approach or is a wrapper the best way to
go?

NB Please tell me to go somewhere else or shut up if I am out of order
asking dumb questions.

Fwd: VS2012 Build and run - notes

Bernhard Wymann — 2012-12-12T16:47:43

Hi all

Here some notes from Steve regarding building in VS2012, thank you very 
much Steve:-)

Best regards

Bernhard

-------- Original Message --------
Subject: VS2012 Build and run - notes
Date: Tue, 11 Dec 2012 13:42:28 -0600
From: SteveO <steveo1720< at >gmail.com>
To: berni4you< at >gmx.ch



Not sure if this is the right place to send the info, or if it already 
exists, but I wanted to share my experiences having now got a building 
running version in VS2012.

1. Make sure the scripts work. For some reason VS creates a file 1.5M 
file ~\Tors\runtimed therefore non of the dev build scripts work cleanly 
as you get a file or directory exists error. Renamed to -old fixed 
issue. Run scripts again.

2. As the Windows libraries were built on an older version, and I do not 
have the source code to rebuild them, you will get multiple errors such as:

Error    16    error LNK2026: module unsafe for SAFESEH image. 
D:\Torcs\src\modules\track\sg.lib(sgd.obj)    track

From
http://msdn.microsoft.com/en-us/library/9a8

Re: Position is (nan, nan, nan)

Bernhard Wymann — 2012-11-30T08:36:32

Hi Jim

Sensors, so do you use 1.3.1 with the competition patch? Which platfrom?

I checked the side right after the start with TORCS 1.3.4, car7-trb1, 
human player, everything seems to be fine.

Beware, car setups are not validated, so if you have a fancy car 
definition file this might let the simulation "explode" (once there is 
somewhere a NAN produced it propagates to whatever it "touches", the 
"things fall apart"), e.g. super strong damping on "fast" values causes 
a NaN somewhere when hitting a high curb on high speed IIRC (e.g. the 
curb of the last chicane before the start on e-track-1, high speed and 
high curb slope and strong damping results in an exorbitant force -> plonk).

There are usually additional compiler switches/pragmas to jump into the 
debugger on FPU exceptions, this way you might be able to catch it for 
investigation, see:
vs2008/10:
http://stackoverflow.com/questions/4454582/visual-studio-c-2008-2010-break-on-float-nan
gcc:
http://www.network-theory.co.uk/docs/gccintro/gccintro_

Re: Position is (nan, nan, nan)

James Ihrig — 2012-11-28T03:21:49

Just realized I may have left out some information that may be important in
that last message:

The track is I am running is "Forza" under "Road Tracks".

The position my car was at was just past the start line somewhere at the
edge of one of the two buildings, (or between them) at the right hand side
of the track.

Jim


On Tue, Nov 27, 2012 at 9:50 PM, James Ihrig <jeihrig< at >gmail.com> wrote:

------------------------------------------------------------------------------
Keep yourself connected to Go Parallel: 
INSIGHTS What's next for parallel hardware, programming and related areas?
Interviews and blogs by thought leaders keep you ahead of the curve.
http://goparallel.sourceforge.net_______________________________________________
Torcs-devel mailing list
Torcs-devel< at >lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/torcs-devel

Position is (nan, nan, nan)

James Ihrig — 2012-11-28T02:50:48

Hello,

I am still running into program crashes once in awhile, but for a different
reason this time. It seemed that my sensors were picking up bad values
sometimes. (I use offset centerline etc. for some calculations.)

My bot currently turns hard right sometimes straight into the wall at the
beginning of the race. Since I am using a neural network this is expected
for some runs. So I added this bit of code into my drive funciton:

static void
drive(int index, tCarElt* car, tSituation *situation)
{
    std::cout << "Positition = ("
        << car->pub.DynGC.pos.x << ", "
        << car->pub.DynGC.pos.y << ", "
        << car->pub.DynGC.pos.z << ") " <<std::endl;
    ...
}

The output is:

...
Positition = (667.472, 1163.4, 3.84071)
Positition = (667.472, 1163.4, 3.84074)
Positition = (667.472, 1163.4, 3.84076)
Positition = (667.472, 1163.4, 3.84079)
Positition = (667.472, 1163.4, 3.84081)
Positition = (667.472, 1163.4, 3.84083)
Positition = (667.472, 1163.4, 3.84084)
Positition = (667.472, 1163.4, 3.84086)

Re: System stability and changing gears in TORCS.

Bernhard Wymann — 2012-11-25T12:51:06

Hi Jim

Ah, ok, the "name" thing is because the "owner" of this is TORCS, so it 
does release the memory at some point, so if you do not "strdup" it 
tries to free "wrong" memory.

situation->currentTime is the current simulation time, yes.

Regarding your "often restart" use case, you should stay with 1.3.4, 
becuase I fixed many (possibly all) remaining memory leaks (see 
changelog), so if you restart very often this can get significant.

I do not know what kind of work you are doing, but there is as well a 
command line mode (e.g. if you want to evolve some AI without watching 
it, see the FAQ, section for researchers) or the blind practice mode 
from the TORCS menu.

Best regards

Bernhard

On 11/25/2012 05:57 AM, James Ihrig wrote:


------------------------------------------------------------------------------
Monitor your physical, virtual and cloud infrastructure from a single
web console. Get in-depth insight into apps, servers, databases, vmware,
SAP, cloud infrastructure, etc. Download 30-day Free

Re: System stability and changing gears in TORCS.

Bernhard Wymann — 2012-11-25T12:34:48

Hi

Yes, if the car damage is more than 10000 damage points it gets removed 
from the race with a "virtual crane" (lift off the track, move to the 
side, let down).

Damage is done in collision.cpp, so you can comment it out/modify it there.

Best regards

Bernhard

On 11/25/2012 06:48 AM, James Ihrig wrote:


------------------------------------------------------------------------------
Monitor your physical, virtual and cloud infrastructure from a single
web console. Get in-depth insight into apps, servers, databases, vmware,
SAP, cloud infrastructure, etc. Download 30-day Free Trial.
Pricing starts from $795 for 25 servers or applications!
http://p.sf.net/sfu/zoho_dev2dev_nov

Re: System stability and changing gears in TORCS.

James Ihrig — 2012-11-25T05:48:07

Timing seems to be better now at real time, (still haven't tested the
simulation while sped up again) I may have compiled the wrong version of my
source. Also, I think I figured out why the vehicle was sometimes float up
and then off the side of the track. It seems to happen when the vehicle
takes heavy damage. I'll just need to look into turning that off.

Jim


On Sat, Nov 24, 2012 at 11:57 PM, James Ihrig <jeihrig< at >gmail.com> wrote:

------------------------------------------------------------------------------
Monitor your physical, virtual and cloud infrastructure from a single
web console. Get in-depth insight into apps, servers, databases, vmware,
SAP, cloud infrastructure, etc. Download 30-day Free Trial.
Pricing starts from $795 for 25 servers or applications!
http://p.sf.net/sfu/zoho_dev2dev_nov_______________________________________________
Torcs-devel mailing list
Torcs-devel< at >lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/torcs-devel

Re: System stability and changing gears in TORCS.

James Ihrig — 2012-11-25T04:57:39

Thank you, I was able to solve both the gear changing problem as well as
the crash on driver selection.

As a note, I think the crash on driver selection may have been related to
my random crashes during runs when I issued a restart. Once I changed my
name and description from:


    modInfo->name    = "MyBot";        /* name of the module (short) */
    modInfo->desc    = "";    /* description of the module (can be long) */

to

    modInfo->name    = strdup("MyBot");        /* name of the module
(short) */
    modInfo->desc    = strdup("");    /* description of the module (can be
long) */

I no longer have an issue with that. I hadn't followed the other driver
format exactly since I am not ever allowing multiple cars in my tests.

I still don't know why my earlier attempts at gear changing didn't work,
but your example code seemed to work. I will look at it more closely on
Monday.

As for the time, I am currently using situation->currentTime, my bot isn't
timing some things correctly but at this point I'm

Re: System stability and changing gears in TORCS.

Bernhard Wymann — 2012-11-24T15:40:12

Hi James

 > I've been running Torcs-1.3.3 and have had a couple problems. First,

I guess this must have something to do with your changes, I have never 
seen/heard of a problem like this before. The problem "makes no sense" 
if you look at how TORCS works: It simulates basically in simulation 
time, just when you have a "watching mode" selected it synchronizes real 
and simulation time to spit out frames, that is it. So I doubt that 
there can be something wrong. I hope you do not rely on the real time 
clock in your robot, only use the simulation time.


No, but it is trivial, the gearbox switches to the gear you choose. How 
you do the decision is then your business;-) But one example is 
explained in the robot tutorial: 
http://www.berniw.org/torcs/robot/ch3/gears.html

This method just works if the gearbox ratios differ and have the correct 
order, so if you sabotage the gearbox you can make this method fail, but 
for any halfway "real" gearbox it works.


Maybe, have a loot at this, this is the only a

System stability and changing gears in TORCS.

James Ihrig — 2012-11-24T04:35:16

Greetings,

I've been running Torcs-1.3.3 and have had a couple problems. First, when
running my bot at a faster speed than real-time, my bot will sometimes
float straight up, then off to the side of the track and the race will end.
(This issue seems like a bug, but I'm reluctant to say so since I've
modified the TORCS code.) Second, I am having trouble with shifting.

I have noticed this code in other bots:
http://codepad.org/bwc5LY4q

I have tried to copy/paste this into my drive() fuction (from other drivers
code) but all I see is the gear bouncing between 1st and 2nd gear, so I
feel I must be missing something. I didn't understand what the code was
doing so I tried a bunch of variations with _gear and _gearCmd and and even
tried manipulating the clutch. Is there any documentation on how gear
changing is supposed to work?

I have tried upgrading to TORCS-1.3.4 to see if my issue with the bot
leaving the course would be fixed, but it crashes once I select a track
when configuring the race. (It seems to be