[robocup-small] Question about the structure of the SSL-Vision packet

[Stefan Zickler]

Dear Jirat and All,

the exact packet format of SSL-Vision's network output has not yet
been finalized. Currently, the system's capture, color-segmentation
and geometry calibration are all in a working state. We are now moving
on to implement pattern detection. The network packet is the final
component to be added. Its exact format will depend on the
feature-set/completeness of the pattern detection component.

Nevertheless, to prepare for integration, we can tell you this much:
We are intending to have the system provide you a list of
locations/certainties of all the robots and the ball. Orientations
should be provided at the very least for the "standard" robot pattern
layout (which is yet to be defined, but likely to be similar to the
"butterfly" multi-dot pattern that many teams use). Additionally you
*might* be able to configure the pattern detector to work with other
teams' patterns in order to obtain their orientations.
If for some unexpected reason, pattern detection will not be completed
in time for competition, then at the very least the system will
provide individual blob information (including color id, centroid +
bounding box locations, pixel count).

For all of the above, the system will be able to provide information
in field coordinates, or it will provide the geometry data and
functions necessary to perform this conversion trivially on the client
side.

As for 2009, the system will most likely NOT provide sensor-fusion. In
other words, it will send you detection results of both cameras
individually, and it will be your team's job to merge them. It is fair
to assume that each frame's packet will be sent with a timestamp value
representing the time of capture.

We are furthermore planning to include a very minimalistic C++ sample
client code which will receive and de-serialize the network packet, so
teams shouldn't have to worry too much about networking.

I am expecting that we will have a more formal definition of the
network packet sometime in mid/late March.
We will announce when SSL-Vision reaches feature completeness and can
be used for testing by all teams. On our original schedule, the
release of the first beta version is scheduled for mid-April, in time
for many of the local competitions. Development so far has been fairly
well on schedule, and we will try our best to meet this date.

Thanks for your patience, and we will keep you updated!
Stefan


On Wed, Feb 11, 2009 at 2:31 AM, Jirat Srisabye <jirat_o at hotmail.com> wrote:
> Dear Stefan Zickler, and the development team
>
> We would like to know the structure of a SSL-Vision packet
> because right now we start coding the vision system.
> We will prepare the global vision code which supports the SSL-Vision packet
> and we would like to test it in GRAZ this year.
>
> Thank you
> SKUBA team
>
> ________________________________
> Windows Live™: Keep your life in sync. See how it works.