Falcons returned from the World Championship RoboCup in Leipzig, Germany, where we put new software and hardware enhancements to the test against the top teams in the world.
One of the most encouraging aspects was how much ASML's robots had increased in reliability. In previous games two or sometimes three robots broke down, but we now ended most games with all five robots in the field. The new, built-from-scratch software had some hiccups during the setup days, but as soon as these were under control it turned out to be very stable and flexible. We now can change robot behaviors or strategy in the halftime breaks without a problem.
However, the results were a bit disappointing: we ended up 8th out of 9 participants. So where did this go wrong? First of all, the participants were a lot more challenging than in the European championship. Second, our software became predictable; where in the European Open we scored occasionally due to random activities from our robots, now the robots did exactly as expected. This was also true in the opponent's point of view, so they could predict what we were going to do and as such block all our attacks. And, we lack speed. The top teams literally drive more than twice as fast.
But then again, perhaps our expectations were a little too high. As team captain Jaap Vos said: "This is a game in which you participate for the long run. You cannot expect to start playing and be a top world player in little over two years. Expectations are that in four or five years we indeed will be one of the top players in this league. There is still a lot to do to get there but I'm confident that we will."
In the end, Leipzig was an amazing experience. We all got to work closely together with ASML colleagues in a challenging environment, and we watched, learned and discussed solutions with the other teams from all over the world. In the end we saw our improvements since the European Open paying off. And it helped to clarify our roadmap on which areas we can improve:
- Tactics: Now we are able to change the behavior of our players in a relatively easy and fast way, we can focus on tactics.
- Speed: This will always be crucial and there will always be the trade-off between speed and accuracy.
- Vision: Our field players currently can see 3m around them. This has to be increased, if possible to the full length of the field.
Now lets start working to get even more awesome robots!