[robocup-small] SSL Vision System!!

[Mani Monajjemi]

Hi All,

We had a discussion about this issue (ID Swapping) between TC members. Here
are the results:

* We will tentatively stick with the current pattern.
* We will add an alternate pattern with three colors (new color is Cyan) to
the ssl-vision trunk in near future, so the teams which have "ID Swapping"
problem can use that in their labs.
* Teams must be ready to switch between these patterns.

Technically, the ID Swapping occurs under high-distortion and gets worse
with bad color calibrations. The solution might simply be to adjust some of
the pattern matcher's weights. The reason why we didn't use Cyan color in
standard pattern was its closeness to blue in the color space which causes
some calibration problems.

With Best Regards,
Mani Monajjemi
On behalf of SSL TC

On Mon, Feb 1, 2010 at 11:53 AM, Yang Liu <yli at puv.fi> wrote:

> Hi,
>
> We have tested SSL-Vision in China Open a few weeks ago and did not
> experience any similar problems. We have run all the games with SSL-Vision,
> except it has crashed once (which was known problem before), the
> identifications of robots, both own and opponent positions and headings,
> were quite stable, actually nothing to complain.
>
> I suppose the ID swapping problem is most probably caused by colour
> calibration, or the hardware (focus of camera, etc.). I don't think the
> lighting will cause any major problem as we use also fluorescent lighting
> in
> our lab. But through careful calibration of colours and adjustment of lens
> focus/shutter speed, this should not be a problem at all. By the way we are
> using standard AVT Stingray cameras with low distortion lens.
>
> Good luck!
>
> Yang Liu
> Botnia
>
> ----- Original Message -----
> From: "Christopher Head" <chead at interchange.ubc.ca>
> To: <robocup-small at cc.gatech.edu>
> Sent: Saturday, January 30, 2010 1:46 AM
> Subject: Re: [robocup-small] SSL Vision System!!
>
>
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
> >
> > It looks like the league standardized on patterns that are identical in
> > rotation except for a very small difference in angle between the front
> > dots and the back dots (specifically, you are seeing IDs 3 and 5). IDs
> > 0 and 4 are also identical in rotation, as are IDs 2 and 6 and 1 and 7.
> > It's a bit disappointing that this decision was made; although a team
> > of vision experts at the competition will certainly be able to
> > configure SSL-Vision to differentiate between these patterns, those of
> > us who are not vision experts and who do not own ultra-high-end cameras
> > or who do not have a testing area with perfect lighting will be faced
> > with significant difficulties setting up testing fields (this is why, at
> > Robocup 2009, our team used patterns with three colours—it is possible
> > to make them unique in all rotations). Jitter and blur was bad enough
> > for us when just detecting colours; detecting angles will make it
> > horrific (please remember that many of us have trouble finding space to
> > set up a field of the required size, and that the areas we can find may
> > be lit by e.g. fluorescent lighting).
> >
> > Organizers: do you have any recommendations? If we test using illegal
> > patterns locally, we may not see these problems only to be faced with
> > them at the competition. If we test using the standard patterns
> > locally, we may spend a lot of time and effort trying to solve these
> > problems, only to discover that they never happen at the competition
> > due to the skill of the vision experts, quality of the cameras, and
> > level of lighting.
> >
> > Thanks,
> > Christopher Head
> > UBC Thunderbots
> >
> > On Fri, 29 Jan 2010 18:07:11 +0800
> > Yusuf Pranggonoh <YUSUFP at SP.EDU.SG> wrote:
> >
> >>
> >> Hi,
> >>
> >> We having a problem with the SSL Shared vision when detecting the ID
> >> of the robot.
> >> The ID of the robot is swaping. We had used the recommended ID
> >> assignment (0-7) as in the rules of SSL 2010.
> >> The attached file is our screen shot of the vision server.
> >>
> >> (See attached file: Screenshot-1.png)
> >>
> >> Is there any setting that we set wrongly?
> >> Your quick advice is appreciated.
> >>
> >> Regards,
> >> Yusuf Pranggonoh
> >> Advanced Robotics & Intelligent Control Centre (ARICC)
> >> Singapore Polytechnic
> >> Telp: +65 6870 8205
> >> Email: yusufp at sp.edu.sg
> >>
> >>
> >>
> >>
> >
> > -----BEGIN PGP SIGNATURE-----
> > Version: GnuPG v2.0.11 (GNU/Linux)
> > Comment: GnuPT 2.7.2
> >
> > iEYEARECAAYFAktjc08ACgkQXUF6hOTGP7eNFgCfS2znduhf197fyLsi3+qInhmE
> > iwAAniWMmKvhejfFdLBbbcSFuX07aiyG
> > =WqnA
> > -----END PGP SIGNATURE-----
> > _______________________________________________
> > robocup-small mailing list
> > robocup-small at cc.gatech.edu
> > https://lists.cc.gatech.edu/mailman/listinfo/robocup-small
> >
>
> _______________________________________________
> robocup-small mailing list
> robocup-small at cc.gatech.edu
> https://lists.cc.gatech.edu/mailman/listinfo/robocup-small
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.cc.gatech.edu/pipermail/robocup-small/attachments/20100206/ffb813ac/attachment.html