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

Hi William

No, the commandline and blind mode do not render the visual output at 
all. Technically it would be possible to achieve this. as long as you 
have an API which allows you to get an OpenGL Context without binding it 
to a window.

What you could try is to start TORCS in normal mode and iconize (minify) 
the window, but I guess the OpenGL Driver/OS will then stop rendering as 
"optimization".

Kind regards

Bernhard

On 24.06.2016 22:56, William wrote:
> Hi,
>
> I am trying to train an agent to drive in TORCS via reinforcement
> learning with visual input (first person view). I was wondering if it's
> possible to have TORCS still render without actually firing up a
> display? (basically still compute everything it would by default show on
> the screen, but not have it actually be shown on the screen...)
>
> Thanks
> William
>
>
> ------------------------------------------------------------------------------
> Attend Shape: An AT&T Tech Expo July 15-16. Meet us at AT&T Park in San
> Francisco, CA to explore cutting-edge tech and listen to tech luminaries
> present their vision of the future. This family event has something for
> everyone, including kids. Get more information and register today.
> http://sdm.link/attshape
>
>
>
> _______________________________________________
> Torcs-users mailing list
> Tor...@li...
> https://lists.sourceforge.net/lists/listinfo/torcs-users
>

Hi,

Here is a patch to torcs-1.3.7 that is applied from outside of
the extracted tarball like this:
patch -p0 < torcs-1.3.7.indismissible.diff
This patch contains a years worth of friction force with rubber
to the road research. This is a call to anyone who wants to
see how a correct vehicle simulator works. The proof is in
the code, and in the driving.

Sincerely,
David Savinkoff

Torcs championship server has support for running Torcs in text mode, 
see section 7.4 here:
https://arxiv.org/pdf/1304.1672.pdf

   Tambet

On 24.06.2016 23:56, William wrote:
> Hi,
>
> I am trying to train an agent to drive in TORCS via reinforcement 
> learning with visual input (first person view). I was wondering if 
> it's possible to have TORCS still render without actually firing up a 
> display? (basically still compute everything it would by default show 
> on the screen, but not have it actually be shown on the screen...)
>
> Thanks
> William
>
>
> ------------------------------------------------------------------------------
> Attend Shape: An AT&T Tech Expo July 15-16. Meet us at AT&T Park in San
> Francisco, CA to explore cutting-edge tech and listen to tech luminaries
> present their vision of the future. This family event has something for
> everyone, including kids. Get more information and register today.
> http://sdm.link/attshape
>
>
> _______________________________________________
> Torcs-users mailing list
> Tor...@li...
> https://lists.sourceforge.net/lists/listinfo/torcs-users

Hi,
I am trying to train an agent to drive in TORCS via reinforcement learning with visual input (first person view). I was wondering if it's possible to have TORCS still render without actually firing up a display? (basically still compute everything it would by default show on the screen, but not have it actually be shown on the screen...)
Thanks William