Today I tried Funpidgin. It's much more fun than plain Pidgin. What I really missed in the latest releases was the ability to resize the text entry. I like it to have more space. It's easier with long entries, you just size it reasonable once and it won't change during input. I also like that the "other side is typing" icon in the upper right corner is back, I don't like the inline text. Why don't the Pidgin developers just add a "Pimp my Pidgin" Tab to preferences with this "special" settings!?
I hope the Pidgin developers get rid of their "we dictate how an IM messenger should work for anybody" attitude and reintegrate the Funpidgin people. Judging from this ticket there are more who would like this... Up to then you can find the package on my fedora webspace. Attention, it conflicts with the Pidgin package, has to since no file is renamed, but just some functionality added.
Wow, when looking in my mail box this morning I was astonished. The president of the RoboCup Federation, Prof. Minoru Asada announced that there will be no RoboCup held in Suzhou this July.
I'm really baffled by this news. That's probably the GAU (acronym for "Größter anzunehmender Unfall", literally: worst imaginable accident) for RoboCup. I'm really curious what kind of solution the trustees will come up with.
My personal interest in this RoboCup besides participating in the RoboCup@Home league is to record evaluation data for my thesis. Such an event is usually a great place to get real-world data and test your stuff - after all this is the primary intention for RoboCup. We'll see...
Responding to Paul's blog entry that we use Fedora on our robots for autonomous robot soccer games and service robotics provoked some echoes. After making it more public what we do on Planet Fedora Jeff Spaleta asked me to start a Robotics SIG - so we do now!
I ask everybody who is interested in robotics to join the effort and bring robotics related software into Fedora and make it fit to power robots, to provide educational tools for robotics starters and to make it a good distro to develop for robots in general.
I have setup a Fedora Robotics SIG wiki page. Please visit and add yourself if you are interested. I have tentatively scheduled an IRC meeting for next Wednesday, May 7th at 15:00 UTC on IRC (see wiki page). Let us know if you want to participate but the time just happens not to be when you are awake so that we can agree on a time where we max the number of participants. If there are enough people interested we can think about creating a fedora-robotics-list.
My personal interest in this is mobile robotics and embodied artificial intelligence. I think Jeff noted some specific interest in the educational sector of robotics. Robotics is a wide field so we need more people involved, maybe even with other specific interest in particluar sub-domains of the field.
The plan is to start identifying interesting software projects that would be worth packaging (and license compatible, always a problem in this field), targeting some robotics-related problems (hardware support comes to my mind) and try to find solutions.
Join now, robotics is a lot of fund and many robots can be built with a small budget already.
RoboCup German Open 2008 is over and we recovered in the last week, filed bugs in our bug tracker, resolved a few issues and noted ideas what to implement and improve for RoboCup 2008 in China.
I have create a new album with photos of the AllemaniACs at the GO2008. It contains various photos from the competition and the robots with comprehensive comments. We are currently preparing the videos for upload to the AllemaniACs' YouTube channel. Up to then I included this year's qualification video in this post, created from own footage by Stefan Schiffer. I love it. You can find the videos in other formats on the qualification page.
Positive in this competition was that there were no bigger hardware flaws. We had loose cables twice which didn't take that long to fix (this time, as they were easily reachable without opening the robot). We had some serious glitches with the Linux kernel however. Most noticeably we couldn't open two different IEEE1394 cameras in two different processes. I have to investigate this. This was still with the "old" firewire stack. The Juju stack gave us an even worse performance (on the mid-size robots). We have very low framerates (because the packet size cannot be set bigger than 4KB) and there is no isochronous transfer bandwidth management or channel allocation. Another problem was that we unplugged the USB sound card while investigating another problem and then it was not automatically recognized anymore when we plugged it back in - and we found this out right at the beginning of the final, thus no sound in the final. Have to investigate this as well.
On the software side we got to the point where we have to admit that we need C++ namespaces in libs to not clash with third party software (Exception is a common class name as we had to learn). A big one on my to do list.
Today is the last day of the RoboCup German Open 2008 competition. We are all absolutely tired, only about 4 hours a night is not enough between quite stressy days.
Yesterday we did a stage 2 test, three times. This allowed us intensive testing and bug fixing and in the end awarded us with 2000 out of 2000 possible points! It was the walk and talk test. In that test the environment is changed just before the test, so that you cannot use a fully pre-built map. The human then guides the robot to 5 selected positions and teaches them (via voice) to the robot. The robot is then ordered to switch into navigation phase and is given commands (via voice again) to go to the just learned locations. We used again the vision and laser based approach described in my last post. That was pretty cool.
Today were the finals. Yesterday we hacked until 2am. Currently we are facing an integration problem between our old and our new software framework that we will have to sort out over the next weeks. We basically pimped our open challenge with a dialog system to have a nicer way to interact with the human via voice and our new face recognition stuff. Just half an hour before the finals when we were testing our robot was going nuts, and we didn't know why. After a few minutes we figured that the battery went dry. This means that the laser scanner does not produce the full 360° scan any more, which confuses the collision avoider and makes it move into strange directions. The localization remained stable though, as it can deal with missing beams, so the problem was not obvious from the beginning. In the investigation of this problem we unplugged the laser scanner and to get to that plug also the USB sound card. After plugging it in again it was not recognized anymore by Alsa and our speech synthesis stopped working, damn! We noticed this right at the beginning of the test, so our human presenter had to fill in the gap in the dialog system. Accidentally it then recognized a person from the audience instead of me in the arena and thus handed the cup it had to fetch over to me, while it should have given it to Vaishak. To compensate we showed our robust collision avoidance by blocking the pathway of the robot multiple times.
In the end we got 49.7 out of 70 possible points. The second team IAIS scored 45.4 and team Homer 33.7. So we were able to defend our title as the winner of the German Open. Whohoo! Now I need to get some food...
(People/robots on the photo left to right top to bottom: Masrur Doostdar, Dr. Alexander Ferrein, Prof. Gerhard Lakemeyer, Ph.D., Stefan Schiffer, Volker Krause, Deniel Beck, Tim Niemueller, Caesar, Vaishak Belle)
Today is the last day of the 
