DailyJournal

22 Juni 2005

18:45 First testmatch of 3 versus 3 (no communication).
1th Attempt the red offensive failed to find its kick-off position.
2nd Attempt all Aibo find their kick-off position. First attack failed by ball out. Red followed, but stopped just before the edge.
18:53 two Aibo's punished for obstruction. One minute later again (two blue this time).
18:54 Possible red goal (difficult to say, goal has no sides)
18:55 Red offensive kicks ball back to own goal. Goalie blocks, Blue attacker kicks, Goalie blocks, Red offensive kicks ball again to own goal (0-1).
18:57 Red pushing goal (four Aibo's in the penalty area): 1-1.
18:59 Blue rush to the goal (unsuccesfull).
19:00 Blue goallie is pushed from the field but find its back quick.
19:01 For the first time a goalie is lost without reason. The goalie leaves the field for a moment, but finds its way back before he is finished.

Because SVN stores the logs on the server, the logs from before the crash of waterhoen are lost (begin May).
svn log works only for a single directory or file, not for all files in the directory.
To perform the count, we should do a ls first, followed by a 'for i in $i'.

Installed cywin's svn-package on pc-aibo.
Able to see logs from the command-line with the command 'svn --username jurgen log'.
Svn gets the logs from the server, instead of storing them locally.
grep -r "| 2005-" returns all the lines with the person who made the commit, which can be used by count_commits.sh.
for i in $(cat ~tmp2_ |gawk '{print $3}' |sort -u); do echo $i $(grep $i ~tmp2_ |wc -l) $(cat ~tmp2_ | wc -l); done >~tmp4_ cat ./~tmp4_ |gawk '{printf("%04d (%02.2f%) %s\n",$2,100*$2/$3,$1)}' |sort -r prints the contribution of each teammember.

Installed http://www.germanteam.org/GT-install.zip on pc-vlab10.
Created domain-users with the command 'mkpasswd -l -d > /etc/passwd' (long list).
Tried to create domain-groups with the command 'mkgroup -l -d > /etc/group', but this failed.
Assigned manually the group 'ii' to 31009.
Changed manually the home-directory of arnoud from //wtcw-server/arnoud to /home/arnoud, which removed the warning of ssh on -lcase.

When the GameControlData is changed from RECEIVER to NET_RECEIVER (which is defined in the UDPhandler.h) the Aibo does NOT listen anymore to game-commands from the RobotControl (as expected). Unfortunally, it also not listens to the GameController2005 (nor on port 3838, nor on port 41392 (A1B0)).

The Dutch Aibo Team played again three games in the round robin; two to reach the quarter-final, and the quarter-final. The latter they just lost with 1-2.
During the night the debuging was switched-off and the number of particles in the self-localisation reduced to 70, and the calibration performed while taking care of the color-distortion inside the Aibo's.
This day showed good soccer play, something to build on.

The Dutch Aibo Team played two games in the round robin, and one practice match, and lost all.
The First game was including a late night remote addition, which resulted in shooting at the goal independent of the color of this goal. Chaotic play.
The second game was with a Floris goallie, and a recent version of the software for the others. Reasonable, but not perfect play.
Before the third game a serious bug was found (the framerate was reduced a factor 4). During the practice match we found out that this was a feature to allow debuging and self-localisation with many particles.

This day was used to extend the calibration from the practice field to the competition field.

Jurgen went to the German Open, to start the calibration as fast as possible.
The inversion matrix of the YCrCb colorspace helped to save and load the colortables.
The testing was difficult because of wlan-problems.