The FUmanoids won the Humanoid KidSize Soccer tournament of German Open 2015!
As a short reminder:
Due to the new set of rules all teams were struggling to play a decent game.
Especially the new artificial turf made walking and getting up hard as never before.
Teaching Alan how to get up.
All our work paid off in the final against Russian newcomers Team Photon from Tomsk.
Even though our first two dribblings failed right before the opponent’s goal, finally we managed to score in overtime.
Have a look at the complete game here (big props to our buddies the Hamburg Bit-Bots, who streamed and uploaded all the games):
The scoring of the goal can be seen at 46:40, the dribbling towards the goal starts approximately at minute 45:00.
A huge Thank You! to the organizers of the event and greetings to all the other teams who made it such great fun.
The Finalists: FUmanoids and Team Photon, Newcomer From Tomsk, Russia.
The morning saw the remaining preliminary games (two per team). After we fixed an issue in the walking code, and added a motion to roll from the front to the back to be able to always get back up, we were quite optimistic that we would score either against the Hamburg Bit-Bots or the Bold Hearts. Unfortunately this bore no fruit. Though the robots had not to be taken out as often after a fall, they still did not hold the ball for a long time.
In the end, all the preliminary games ended 0:0, making it impossible to rank the teams in the group phase. The rule book has a provision for that case, saying that each team takes alternating penalty kick trials trying to score a goal. As this forbids entering the penalty area and no team is currently able to kick the ball over that distance (or at all), a new approach was decided upon. Five alternating trials would be hold, the first being a regular penalty kick, the second a modified penalty kick where the striker may enter the penalty area, the third trial would be judged based on the distance the ball is moved towards the goal, the fourth trial on the distance walked towards the goal and, if required, the fifth trial would be won by the team that walked the farthest (whatever direction).
In preparation of this phase, a check of our code revealed a nasty bug that we added yesterday. This turned out to be the reason the robot would walk away from the ball. Having fixed that, we went into the penalty trials quite optimistically. As expected, the first trial was no success as we can’t kick. However in the second trial we scored our first goal:
Before us, Photon also managed to score and so did Bold Hearts. Ranked on time, this made us first in group, followed by Photon and Bold Hearts. The fourth in the group turned out to be WF Wolves after they managed to move the ball farther towards the goal than the Hamburg Bit-Bots in the firth trial round. This meant we would be meeting WF Wolves in the semi-final.
In the semi-finals we started playing in the “bad” direction, where the grass’ “growing” direction makes it much harder to walk. This caused a lot of falls and ultimately no goal. However in the second half, playing with the grain, we finally managed to score. Our work on walking and stabilization payed out when we withstood direct contact with the large WF Wolves robot and Grace dribbled the ball close to the goal. When she lost the ball, Alan took over and finished the way into the Wolves’ goal.
In the other semi-final, Photon won against Bold Hearts by a technical goal, dribbling the ball 12cm farther under the modified penalty kick rules. We will meet Photon tomorrow in the finals.
Today were the first half of the preliminaries (round robin). We played two games, which you can see below in the recorded live-streams the Hamburg Bit-Bots provided.
Alan taking possession of the ball.
We tied 0:0 in both games. However we managed to gain possession of the ball quite often and move it toward the opponent goal. In both games (as well as the other games played today) we were the only team to manage this in that extent. Most teams had severe problems getting either back up or walking more than a step or two, whereas our robots managed to recognize the ball and walk over longer distances. Unfortunately this was not quite enough to score a goal, as we were still plagued from falls and especially our on-going inability to stand up when having fallen on the front. The latter required the robots to be taken out, which cost us valuable time. It was also unfortunate that the robots “lost” the ball a few times when close to the goal and wandered off to somewhere else.
We are looking forward to the games tomorrow (10:15 against Hamburg Bit-Bots, and 12:15 against Bold Hearts). We have started added further improvements which we hope will allow us to score a goal tomorrow. Stay tuned, come by to cheer us on or at the least watch it live (thanks to the great service by the Hamburg Bit-Bots).
Yesterday was the second and last setup day. We spent a lot of time testing and improving the robot software in the morning, in anticipation of a friendly game against the Hamburg Bit-Bots in the afternoon. Progress was good, we managed to get the walking more stable and standing up at least from the back (so far). After lunch and a referee training, it suddenly was 4:30pm and we played almost a full test game against Hamburg Bit-Bots.
As test games are designed to do, this showed us quite a few problems and what we would have to work on for the rest of the day. We suddenly had immense problems gaining any ground on the grass and a last-minute bug in the code caused most of the robots to just refuse to do any work at all. The Bit-Bots didn’t score much better and spent most of the game lying on the ground. We also had problems getting up, mostly when falling on the front.
The software crash was quickly fixed after the game, and the walking algorithm also received additional attention (though as it was working much better before and after the test game, it may have simply been robot stage-fright). In addition to the artificial grass itself, additional challenges arose by the way the grass turf was being laid down, having a few bumps and a small trench at the part where the two halves meet. After some fine-tuning it looks a bit better, but it will still be interesting to see how we (and the other teams) will handle it.
Today (Friday) the games are starting. The schedule is up at http://www.robocupgermanopen.de/schedule/major/humanoid-kid-size, we will be playing our first game at 2:15pm against the WF Wolves, and the second one at 4:15pm against the new team Photon. A live-stream by the Bit-Bots will hopefully allow you to see the game live (we will post the link to our Facebook page – which is visible without a Facebook account, by the way).
Yesterday morning (Wednesday) we arrived in the lab early in the morning and drove to RoboCup German Open in Magdeburg. After arriving and unloading the cars, the first task was to inspect the playing field. As you know this year’s rules require a new, more challenging ball (at least 50% white, not the uniform orange color from the previous years) and new all-white goals (previously yellow) which we could already test at the Iran Open earlier this month. The third big change this year is the artificial grass. At Iran Open the turf was easily manageable, however in Magdeburg they picked a challenging one. Certified for FIFA games, this turf will require a lot of new skills from the robots of all teams.
The new artificial grass field at German Open
At the end of the day, we have verified and slightly improved our robot software for the new tasks. Our new ball detection algorithm works remarkably well and we seem to see the ball (at least for now) much better than most of the other teams. We also tested our updated walking algorithm, and are confident that we will be able to handle the high and challenging grass in that regards. Unfortunately, and quite surprisingly, the robot motions for standing back up after a fall stopped working. It worked on the comparatively similar turf we tried back home, so this came quite as a shock (incidentally many of the other teams have/had similar problems). This is something we will have to work on today.
Alan relaxing on the grass.
Emmy checking out the goals at German Open
We don’t know the game schedule yet, as only five out of six teams have arrived. Depending on whether the last team will come to German Open or not, the preliminaries will either be split into two 3-team groups (with each team having to play 2 games) or a large 5-team group (requiring each team to play 4 games). It seems games will start Friday afternoon either way and continue on Saturday, with the semi-finals in the afternoon.
P.S.: Starting Friday morning, RoboCup German Open will be open (with free entrance) to the public.
One thing worth pointing out how well the players communicated with another. The goalie Grace for example told Alan that he was dribbling towards the wrong goal, which allowed him to reorientate himself.
The first game against the Hamburg BitBots was exciting as always, both teams had improved quite a bit since the testing match on Tuesday.
One of their defenders was nice enough to help the ball over the goal line when we had only made it up to it.
After more great chances unfortunately not taken, in the end we won with a score of 1:0.
Our last game of the groupphase against our friendly table-neighbours of the team MRL ended 1:0 as well.
The redeeming goal needed a painfully long time to happen, and we were close to shooting at our own goal before that.
A sped up walker gave us more ballcontrol in the semifinals, in which we met with the BitBots again.
The game was quite the thriller, and the result (1:0) shows how close we were to a tie.
After this exciting day of Football and Robots and nice people and interviews with Iranian newspapers and TV stations and ice cream and exploding batteries (well, there were sparks..) we are all eager to play against Parand again in the final tomorrow.
This rematch of the tie on wednesday will be an interesting game for sure, we will let you know how it goes.
Wish us luck!
The friendly match on Tuesday against our friends, the Hamburg BitBots, provided us with enough material to be busy coding a long way into the night.
But the hassle paid off:
The next morning, in our first game against the Iranian team Parand, we still encountered problems, but a lot less.
The Parand strikers were strong shooters, but the FUmanoids prevented any clear goal, so ultimately, the game ended in a tie (0:0).
Having learned from the first match, our second one against the team AMITIS turned out to be more successful:
The Amazing Alan scored the first goal in the competition at all by dribbling it over the line. What a success!
The new rules apparently prove difficult for all teams.
Everyone seems to have trouble finding both ball and goal.
So it keeps being interesting to see how the competition will continue.
Oh right: Lutz held a presentation on one of his favorite topics: Inverse Kinematics.
It was both bold and beautiful, believe me.
Happy Easter, dear Fans and fellow FUmanoids!
Our easter-weekend was pretty busy: filled with preparations and teambuilding exercises in beautiful surroundings (look at those mountains!).
Yesterday was the first setup-day on location and the first chance to test the robots on the playing field while around us the exhibition was being prepared.
Today the IranOpen 2015 was officially opened with a ceremony including many important-looking people, participants from many different countries and an indoor fireworks.
Some of our achievements over the last few days are a temperature-dependent gyrodrift, a quicker localization and improved (no, brillantly working!) ballrecognition (new rules replaced the former orange ball by a more complex one including white and arbitrary colors).
A huge relief is the artificial grass on the field: It is much shorter than expected and the ground is pretty sound, so it is relatively easy for the robots to walk on it.
A list of our games for the next two days:
Wednesday 10 am: Parand (Iran)
Wednesday 2 pm: AMITIS (Iran)
Thursday 10 am: Hamburg BitBots (Germany)
Thursday 3 pm: MRL (Iran)