VISION RoboCup-98

Sept. 30 - Oct. 1, 1998, VISION trade fair, Stuttgart


The VISION RoboCup-98 took place from Sept. 30 until Oct. 1, 1998 at the VISION trade fair (the largest industrial vision trade fair in Europe) at Stuttgart. It was held as a national (Open German) competition following the rules of the  Robot World Cup Initiative (RoboCup). The VISION RoboCup-98 was organized by Prof. A. Zell (Univ. Tübingen) with the help of the Messe AG, Stuttgart.

Teams and Robots

All European teams (that we knew of from RoboCup-98 in Paris) were invited. Four German middle size teams had confirmed their attendance. They were:
  • CS Freiburg (J.-S. Gutmann, W. Hatzack, I. Herrmann, B. Nebel, F. Rittinger, A. Topor, T. Weigel, B. Welsch): 5 Pioneer1 with Sick PLS200 laser scanner, Newton Lab vision system, WaveLan radio ethernet, Libretto notebooks, kick mechanism with solenoids and Märklin metal construction set, ball guidance with spring mounted fingers, goalkeeper has laser scanner, vision system and kicker mounted sideways. One-shot kicker for penalty shooting (could hardly be wound up by one person)
  • T-Team (now Attempto Tübingen) (T-Team M. Plagge, D. Jung, R. Günther, S. Fleck, A. Zell): 4 Pioneer1 robots with Pentium200 resp. AMD K6 onboard-PCs, 64 MB RAM (incl. RAMDisk) and Sony pan/tilt camera (pan/tilt feature not used), custom spring-based translational kick mechanism with electrical wind-up, ball guiding springs on front sides, 2 Mb/s ARtem radio ethernet, one Pioneer AT robot with Newton Lab vision system as goalkeeper.
  • Ulm Sparrows (G. Kraetzschmar, S. Enderle et al.): 3 Pioneer1 robots with vertical air pressure tanks  (sponsored by Festo, Inc.), each with onboard-PC based on PC-104 and PC-104 frame grabber, a tracked vehicle with kicker which moved parallel in front of the goal line.
  • Agilo RoboCuppers, TU Munich (M. Klupsch, T. Bandlow, M. Grimme, I. Kellerer, M. Lückenhaus, F. Schwarzer, C. Zierl): 5 Pioneer1 robots with Lanner  EM500 single-board computer (Pentium 200 MHz MMX, 64MB RAM, onboard ethernet, Videologic PCI video capture card), remote master PC, linked with WaveCell radio ethernet, vision system based on commercial HALCON system.
All teams played with their robots demonstrated at Paris, with the following  improvements:
  • The Univ. of Freiburg had the chance to demonstrate their one-shot penalty shooting catapult (not used in Paris, but used here against Tübingen with great success).
  • The Univ. of Tübingen tested their new spring-based kickers (also with great success against Munich).
  • The Univ. of Ulm had modified the heads of their robots and had significantly improved their reliability.
  • The Univ. of Munich had improved the robustness of their hardware and software.
Although all teams used the Pioneer 1 robots by ActivMedia as base platform (Tübingen also a Pioneer AT as goalkeeper), it was interesting to see how the robots differed.

Rules

Each team played against each team (round robin). The field size and colors and most of the rules were the same as in Paris, with the exception that we tested the proposition for RoboCup-99 to play 10 min. in each half. This resulted in more goals per game, more fun for the spectators, and improved chances for better teams to score against less capable teams. No team ran out of batteries, although some (relatively few) robots became immobile due to software problems. Thus, this rule change seems very advisable for RoboCup-99 at Stockholm.

There was also a minor change in penalty kicks: we placed the ball halfway between field center point and goal line and started the robots from some place in the other half (chosen by the shooting team, field center point, if desired). This was more interesting than kicking directly from behind the ball. This minor change can also be recommended.
With the improved kickers and longer game only two games went into penalty shooting and none into the physical challenge round.

We did not change the size of the goals (as proposed for Stockholm), as we used a playing field which had been built at Ulm to prepare for Paris.

We also allowed the controversial ball guiding spring fingers which gave better ball manipulation, better ball recovery at walls and fewer ball losses when moving forward with the ball but also resulted in relatively frequent "finger hooking". In most of these cases the robots could separate themselves after struggling a bit. Freiburg, Tübingen and Ulm used these fingers, Munich had a concave metal front which was only slightly less effective but also had similar problems.

Schedule and Results

The schedule was designed to minimize conflicts with sessions of the DAGM-98 (German Pattern Recognition Society) and the special AMS (vision cars) session. Thus, each game had a big crowd of spectators around the field. Besides, attendance constraints (Freiburg wanted to attend on Oct. 1 only) had to be taken into account.
 
Wednesday, Sept. 30, 1998
Time Game Result Points
10:30 - 11:30 Tübingen - Ulm 0:1 0:3
12:30 - 13:30 Ulm - Munich 2:1 3:0
16:00 - 17:00 Tübingen - Munich 1:1 / 4:1 (penalty) 3:0

Comments:

  • In the game Tübingen - Ulm, Ulm scored against Tübingen in the first half which had problems with their Pioneer AT goalkeeper (going up the goal posts). In the second half Tübingen moved the ball up to the goal line of Ulm but not beyond it.
  • In the game Ulm - Munich, Munich scored two goals (unfortunately one into their own goal) so that they lost the game.
  • Both Tübingen and Munich scored a proper goal in each half, Tübingen with a beautifully long shot with their new kickers, necessitating a penalty shooting. Here, much to the surprise of the Tübingen team itself, their attacker scored three penalty shots while the goalkeeper held all acceleration shots from Munich which lacked a kicker.
 
Thursday, Oct.1, 1998
Time Game Result Points
10:30 - 11:30 Ulm - Freiburg 0:4 0:3
12:30 - 13:30 Freiburg - Tübingen 0:0 / 4:1 (penalty) 3:0
14:30 - 15:30 Munich - Freiburg 0:2 0:3

Comments

  • In the fourth game Freiburg showed their superiority by scoring two field goals in each half against Ulm.
  • The rematch of the final at Paris of Freiburg against Tübingen was a very lucky draw for Tübingen in the regular playing time: three times a robot of Freiburg moved the ball directly in front of the goal line of Tübingen but each time the attacker then fancied it had done enough already and returned to its starting position. Then the goalkeeper for Freiburg only permitted one penalty goal for Tübingen, while the Freiburg catapult-like one-shot kicker (based on a luggage fixing string with a solenoid trigger) scored 4 penalty goals together with their more precise aiming.
  • In the last game Freiburg again was superior and scored two goals thanks to their powerful absolute positioning system. Munich had problems because their defenders frequently got stuck in their own goal area and had to be removed.
 
VISION RoboCup-98, Final Scores
 
Freiburg
Tübingen
Ulm
Munich
Sum
Place
Freiburg
XXXXXXX
3
3
3
9
1.
Tübingen
0
XXXXXXX
0
3
3
3.
Ulm
0
3
XXXXXXX
3
6
2.
Munich
0
3
3
XXXXXXX
0
4.

Prof. B. Nebel from the Univ. of Freiburg was presented the VISION RoboCup-98 by the organizers.

Lessons learned about the robot teams

Lighting and wall colors

Seeing the ball (and only the ball as ball!) was again the most difficult challenge for the teams. Two Halogen lamps hanging down from the ceiling provided a relatively homogeneous, but unfortunately rather reddish light which troubled the Newton Lab system. Thus, robots were again distracted sometimes by reflections on the wall.

Global positioning

The Freiburg team with its global positioning system based on laser range finders still has to be beaten. Ulm and Tübingen only had a local visually based positioning, and Munich had a visual 3d-model matching but with lower frequency.

Fingers and Robot Size

Spring fingers were effective but in some cases also hindered robots. As their presence or absence changes the range of potentially successful robot maneuvers significantly the questions whether fingers are permitted and the definition of the size and shape of the robots should be decided rather quickly now.

Kicking device

With many other things being similar, the lack of a good kicking device hampered the performance of Munich in the regular games. In each penalty shooting the team with the stronger kicking device won (as the reaction time for the goalkeeper goes to zero). It should be considered whether one-shot "canons" or "catapults" should be allowed at Stockholm.

Reliability

Reliability of the robots still is a big issue. The team of Tübingen had a power supply breakdown in two robots (one could be replaced before the games) and camera/cable problems. The new powerful kicking device had a trigger bar which often made the ball slide instead of move by accident which might have provoked protests on other terms and conditions.

Field size and number of robots

The current field size is too small for 5 robots of each team. As enlarging the field seems to be impossible at Stockholm the proposal to play with 4 robots instead of 5 (which I rejected initially) should be reconsidered.

Images of the Games

The images (scanned from prints of slides) are of low quality because flashes were prohibited.
 
a) Tübingen against Munich in play
b) dito, but different game situation
c) Munich scoring the 1:1 against Tübingen
d) the crowd watching the game
e) the winning team of Freiburg (again)