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

Greetings,

Using the latest TORCS CVS, I get an absolutely reliable segfault
every time I try and configure a race. I can scroll through the tracks
and select one, then when I press <ACCEPT> I get the following message
delivered by 'torcs -d':


Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread -1227290912 (LWP 12182)]
GfParmGetStr (parmHandle=3D0x0, path=3D0xb7f15b40 "Robots/index/1",
    key=3D0xb7f1349d "", deflt=3D0xb7f1349d "") at params.cpp:1913
1913            if (handle->magic !=3D PARM_MAGIC) {
(gdb) #0  GfParmGetStr (parmHandle=3D0x0, path=3D0xb7f15b40 "Robots/index/1=
",
    key=3D0xb7f1349d "", deflt=3D0xb7f1349d "") at params.cpp:1913
#1  0xb7f0f3ea in RmDriversSelect (vs=3D0x0) at driverselect.cpp:335
#2  0xb7e90181 in reConfigRunState () at racemanmenu.cpp:159
#3  0xb7edde39 in GfuiScreenActivate (screen=3D0x0) at gui.cpp:464
#4  0xb7f0ddfd in rmtsDeactivate (screen=3D0x844c418) at trackselect.cpp:82
#5  0xb7ee0f95 in gfuiButtonAction (action=3D-1208929123) at guibutton.cpp:=
577
#6  0xb7edfd9c in gfuiMouseAction (vaction=3D0xb7f15b40) at guiobject.cpp:4=
58
#7  0xb7eddae6 in gfuiMouse (button=3D0, state=3D1, x=3D137600, y=3D423) at=
 gui.cpp:396
#8  0xb7a91c1a in glutMainLoopEvent () at freeglut_main.c:1234
#9  0xb7a91e85 in glutMainLoop () at freeglut_main.c:1489
#10 0x08048e55 in main (argc=3D7, argv=3D0xbfa2c634) at main.cpp:110

I don't get any problem with modifying drivers or actually racing, I
just can't choose which driver to use, or indeed track. Freeglut is
from CVS, incidentally, which is purely to avoid a problem that kills
Flightgear with current Freeglut.

Cheers,
JH

A minor update on CVS, removing the dependency of road noise to rolling
resistance (now it depends only on speed and reaction force) and adding
slight tweaks in volume. Rougher tracks should still sound somewhat
louder if only because of the random increases in reaction force.
-- 
Christos Dimitrakakis
Homepage: http://www.idiap.ch/~dimitrak/main.html
Music: http://olethros.dmusic.com

I have an update on the control tutorial, with a new section on
trajectory optimisation - this might be useful for your robot.

http://www.idiap.ch/~dimitrak/simu/index.html

here is the source for the demo:

http://www.idiap.ch/~dimitrak/software/trajectory/optimise_road.cc

After compling, run it with

optimise_road 1000 0.01 >trajectory.out 2>learning.out

The trajectory.out file will be split into 3 pairs of 2 columns. One
pair is the x,y coordinates of the track's left side, the other the
track's right side, and the 3rd the coordinates of the trajectory.

The learning.out file will haeve the following columns
iteration, cost, penalty, mean squared derivative, mean derivative (the
last one is not very meaningful mathematically in terms of convergence
though).

If the penalty remains at 0 then the algorithm probably has not
converged, because most of the optimal solutions tend to touch the sides
of the road.

OK, have fun


-- 
Christos Dimitrakakis
Homepage: http://www.idiap.ch/~dimitrak/main.html
Music: http://olethros.dmusic.com