After being briefly introduced and tried out in RoboCup 2016, this year officially included the Drop-In Games. Four teams bring one robot each to a joint team. This encourages teamplay between teams as well as cooperation. These games are especially challenging for the referees as they have to look out for eight robot handlers (one for each robot). For the teams it is interesting to see which team has which strength. We believe that you can get a good overview of the league through these games. The results are scored in a separate scoring system. Each robot gets points for being in play while a goal is scored and more points for scoring a goal themselves.

 

 

 

 

 

 

 

 

 

 

The regular games started with a preliminary that determines the composition of the round of sixteen.

Yesterday we played SEU-UniRobot from China (loosing 0:1 to them) and ITAndroids from Brazil (ending in a 0:0). Today we played MRL-HSL from Iran loosing with a 0:1. We now stand at 3rd place in our group and we will play the NUBots from Australia tomorrow.

 

 

 

 

 

 

 

 

 

Today we were able to use our “new” robot for the first time. In actuality it is a Minibot with newly manufactured parts for the legs and arms and an old torso. The missing metal parts for the torso which we replaced with old parts were meant to arrive at our lab about a week before we took our flights to Japan but due to difficulties at the post office we never received them and they were sent back.  After assembling him here at the venue we got him through robot inspection and he even received a name: We call him Frankenbot.

 

 

Furthermore we used our time to exchange with other teams about different robot platforms, since we are looking for a new one ourselves. New materials such as carbon fiber are especially interesting to us.

It was very relaxing to get some sleep over night in a comfy bed and after a short breakfast (cereals with milk), we started our way to the competition halls.
Travelling by train gets easier each time, the biggest difficulty is to find the correct price for the station you want to travel to. Then you just buy a ticket with the corresponding price and use it to enter and leave the train system. For the next day it will be even easier as the organisers will be handing out train tickets to RoboCup participants so that we don’t have to queue at the ticket machines anymore.

Arriving at the venue we quickly saw that the organizers prepared a lot to avoid mistakes of some previous RoboCups. The registration area was clearly laid-out and it took us only a few minutes to get our team badges and train vouchers. We found the competition fields to be fully prepared, too, and quickly started to set up our team tables.

Next on our schedule:

• Calibration of the camera to adapt to the light in the hall
• Making sure the robots survived the travel fine and are able to start the system
• Testing the walking
• Robot Inspection by the Technical Committee

It seems like yesterday that RoboCup 2016 was in Germany, the world championship on home soil – and suddenly the next world championship is just around the corner! In 2017 it takes place in Nagoya, Japan. While Maike was the first to arrive in Nagoya already on Saturday, almost all other team members finally arrived today as well. Tomorrow, just in time for the start of the two set-up days, Jasper will be the last of our team to join us and will thereby reunite our team again.

After a total travel time of more than 12 hours, first to Helsinki and then to Nagoya (for Daniel and Niklas with an additional stopover in Tokyo), everyone is totally tired. Nevertheless, we still had enough energy to explore the surroundings of  our hostel near the central station in Nagoya at least a little bit. As a member of the organizing committee, Maike was already able to inspect the hall for the very first time. All in all, it looks very promising, even if our robots will likely have some problems dealing with some gaps in the artificial turf. However, we will only be able to investigate this more tomorrow, when we are all permitted to enter the hall with our whole equipment.

After a traditional Japanese dinner, the suitcases are now packed with the equipment for tomorrow and we are looking forward to starting a new RoboCup adventure at 7:30 Japanese time (which is 0:30 German time)!

Since the beginning of 2017 we are heavily working on our new software architecture. We decided to move away from our old self-written system. Starting with the competitions this year, the RoboCup GermanOpen and IranOpen, our robots are using ROS (see ros.org). This gives us lots of benefits but also there is still a lot of work to do.

Most of our code is rewritten or integrated into ROS by now. Since we can use the same programming languages as before (Python and C++), we were able to keep parts of our old codebase. We are going this step together with other teams from the humanoid kid-size league who also are using ROS and the proposed architecture. This enables exchangeability of the ROS nodes. On this way, we were already capable of testing the other team’s vision code with minimal effort on our system.

Other benefits are the debugging tools and other libraries provided by ROS.

Our software is open source, you can find it on https://github.com/bit-bots/.

We are heavily encouraging other teams also to use our proposed architecture or getting in touch with us to discuss further improvements.

The new architecture was developed as a master’s thesis (see Publications) and will be presented at the RoboCup Symposium 2017 in Nagoya, Japan. We hope that an open architecture can help other teams to focus on improving their codebase and ease the access to the league for new teams.

Today is already the last day of the German Open. Here a short overview of what we did the last days.

 

We made sure the software was running correctly on all robots so that they are able to play. We worked on Minibot’s walking, and transferred existing motion sequences of Minibot standing up to the new system. Also we put plates under Minibot’s feet, to expand the foot area.

The vision continued working (here a small correction: our vision was not made faster by simply removing features that we don’t need but through algorithm optimisation). We can now change or adapt parameters over wlan while the robot is running

(Black circles are balls that are disregarded, since they are over the horizon, the red line.)

 

Parts of our working group were busy debugging and revising the GameController. We also worked on a tool to visualise what the robot thinks or knows about the field, like its localisation and where it assumes the ball to be.

 

Having just one Minibot-robot posed a significant problem in the last days . Since software engineering is focused on the future and thus on software for the Minibot-platform, we need Minibot for lots of testing. Since we worked on vision, behavior, walking and standing-up, several people needed to wait for their turn.

We had planned to already have a second Minibot by now, but due to delays in production sequence we don’t have it yet. In the future it will be essential to have a second (better even a third) Minibot to not delay working on code or testing.

The article about the games from yesterday will hopefully follow shortly after this one.

Yesterday the first two games took place: Drop-In Games. Two teams play together against the other two teams on the German Open, and each team supplies two robots.

Around 4 pm the first game started, in which the Rhoban Football Club play together with the WF Wolves against the 1. RFC Berlin (formerly FUmanoids) and the Hamburg Bit-Bots.

 

The game at 7 pm took place with the same combination of teams. In neither game a goal was scored, but robots from both team Rhoban and RFC Berlin saw the ball and had contact with the ball several times.

 

This morning the third game took place, in which the RFC Berlin played with the WF Wolves against Rhoban Football Club and Hamburg Bit-Bots.

One of our robots, Wilma, managed to see the ball at Kickoff and walked toward it,  but fell. Before she managed to fully get up, two other robots, one from Rhoban and one from Berlin reached the ball, but stood too close to each other so that neither could properly kick the ball.

The robot from Rhoban moved the ball a bit but then fell on it. A second robot from team Rhoban reached the ball and began to manoeuvre the ball into a corner next to their own goal, and to repeatedly move the ball out of the playing field.

 

Wilma tried to join the game after a while, and managed to slowly but surely get up, but she fell when she tried to start walking. One robot of the Wolves ran into a goalpost and one robot of team Rhoban strayed into the opponent’s goal.

The ball got moved to the middle line, and the robot from team Rhoban walked straight towards it, and past it, just to manoeuvre the ball into the same corner again where it went out of the playing field again.

With that, the game also ended in a tie.

 

Time really flies during competitions. Here is just a short recap of what we all did yesterday.

Since the vision of our robots was rather slow (it worked, but it was too slow to be useful in a game), we used the vision of the WF Wolves, and adapted it to our robots and our needs (for example they also factor in how the head of the robot is inclined; we don’t have this data). In general we threw out a few features to make the vision faster.

The colour config got adapted to this surrounding and artificial turf. We are still working on computing the horizon, currently it starts from below which results in a wrong horizon so that balls get left out. We will try to compute it from above.

Furthermore we are working on a ‘top-down’ object detection, this means we know what we are looking for in a picture (the ball). This method is based on human perception.

 

For some time we have been using ROS, Robot Operating System, which provides a consistent standard (and made it possible to just use the vision of the WF Wolves). ROS has many advantages, but our RecordUI, with which we recorded movement sequences, does not work anymore. That’s why a few people sat together and wrote a plugin for ROS so that we can record sequences again.

 

Also we wrote a boot-script, so that software that is already on the robot will be executed when started, even without connection to a computer. The walking-configuration was adjusted to the really short artificial turf here. Motors in Minibot which were older and used were changed. Several things, especially the flight to Nagoya, were organised for the World Championship. Older and out-of-date Plugins for this website were updated or changed so that this article hopefully will be linked on Facebook and Twitter again.

Two team members were occupied with organisational stuff like game plans for the German Open; for example we will have several Drop-In Games next to the normal games. Here this will mean that two teams play together, and each has to provide two robots and play against the other two teams with two robots each. The GameController got fixed and the teams were inscribed.

After this exhausting but productive day we had roasted potatoes with red cabbage, as well as spaghetti with tomato sauce for dinner, so that hopefully we could continue to concentrate on working today.

 

Today is the first day in the hall. After breakfast at around 7:30 the first group departed, since the hall opened at 8 am. Half an hour later the rest followed.

The shelf we brought along was assembled quickly and the most important things, our food supplies, were put away.

 

After setting up the network, connecting monitors to laptops and plugging in the laptops, a period of quiet working began – although it is rather loud in the hall itself, many teams are still arriving and unpacking, and there are a number of things that have to be set up for the competition.

We started adapting the ball detection for the field and getting minibot to walk on the relatively short artificial turf. Probably we will also put the old Darwins to use.

Today we met at 9 o’ clock to pack for our trip to the German Open in Magdeburg. This time we had vans with nine seats and lots of storage room, so taking everything with us was no problem.

At 11:30 we were finished loading everything in and after lunch we headed to Magdeburg.

We have three smaller vacation apartments and are located directly next to a supermarket where we bought groceries for this week.

Peeling potatoes, cutting carrots, cucumbers and bell peppers, putting potatoe wedges, spring rolls and chicken nuggets in the oven, some salat, and dinner with something for everyone was ready.

 

Afterwards we had a team meeting where we talked about our plans and assignments on this competition.

Tomorrow morning we will look for a good spot in the fair hall and will update you then.

Unfortunately, as we had not managed to score a goal in the second group stage, this was the end of this year’s Iran Open for us.

We were able to trace the problem of falling over while running to the fact that the gearboxes of some leg motors were worn out and that the parameters of the walking algorithm were not well adapted to the new robot. Despite the relatively difficult conditions due to the conversion of the software to ROS, the competition still was a successful start in the season and we are looking forward to the German Open at the beginning of May in Magdeburg.

The FUmanoids won the game for the third place against Robit. The final was played between MRL and the ZJU Dancers and was an exciting match. The spectators supported strongly the local team from Iran. Unfortunately there were problems with the the game controller, the software which tells the robots the current status of the game. After the half-time pause, the status was not set to “play” again, so some robots did not move. This error was not noticed for several minutes, so it was added a big overtime. As a result, MRL was able to shorten the 4-1 lead to a 4-2 lead. But against the runner-up world champion, this is probably not a shame, although the spectators were very disappointed.

Between the games, we used the time to do some networking with other teams,  helped quickly repair other teams’ robots and visited the exhibits of the aligning university. In the middle-size league, a robot team played a test game against a few volunteering boys from the audience.

At the closing event in the evening at 6 p.m. unfortunately we arrived 10 minutes late and were a bit surprised that it had already begun. Nevertheless, we took our honorary seats in the first rows. Unfortunately, none of us got a translation device, so that the speeches were somewhat monotonous. This year, there were separate ceremonies for the junior and major leagues, so the event lasted only an hour. Afterwards the banquet took place in the competition hall. We met different teams, took some selfies with them and exchanged email addresses.

In the evening we decided to visit the nearby Tajrish Square. At the front desk of the hotel we ordered two taxis. The about 20-minute trip costs about 3€ per taxi. There we strolled over the closing bazaar, which some of us already knew from the previous years and the adjoining mosque. The trip back to the hotel turned out to be more exciting: we negotiated with a taxi driver a plausible price and showed him a letter of our hotel, which explained the way there. The car did not seem to pass through the German TÜV test, but since the top speed due to the traffic situation in Tehran seems to be about 20 km/ h, we were not worried. One of the two taxi drivers seemed to know the way and the other wanted to follow him. Unfortunately they lost themselves in the traffic. After the taxi driver, of whom we had thought he knew the way, had asked other people at the roadside  several times, we actually reached the destination. And our friends in the other taxi arrived a few minutes later. We spent a nice remaining evening at the hotel with our interpreter Mahdi.

The next day was completely reserved for sightseeing. Since we should be in a larger group, we had to pick another team at the exhibition hall. Since we were no longer with buses of the university, we could not simply go over the exhibition grounds to the meeting place. The route outside was about 20km long and the journey lasted half an hour. Walking we would have been there in 15 minutes. We started our tour at the Golestan Palace complex, which is really very worth seeing. It is located close to the large bazaar of Tehran. There we split up into smaller groups, lunched a delicious Iranian burger and covered with spices, cloths and other souvenirs.

Afterwards we went on to the Nature Bridge, which connects two parks across a large road and houses even cafes and viewing platforms. The actually relatively short way to get there was extended by a traffic jam for a considerable time. Many of us used the opportunity to make a little sleep in the more or less comfortable bus to catch up. Arrived at one of the connected parks we made a pleasant walk and enjoyed the view from the bridge.

After dark we drove to the Milad Tower, the television tower, and enjoyed the spectacular view. A in theory successful finish of the tour, only the trip back to the hotel was problemfull. The bus driver unfortunately drove 20 minutes into the opposite direction of our hotel. He had to go back, of course. With some help from other car drivers we arrived finally.

The team ER-Force from Erlangen was already waiting for us, because we wanted to go to the airport together with another, bigger bus and their flight was a bit earlier than ours and were promised a bus even earlier. We had not eaten since noon and so we had to take the food from the hotel restaurant into the bus.

Arriving at the airport we had to wait some time until we could check in. Also when the information displays would let us believe we could get rid of our suitcases, the check-in desk was still unoccupied for a felt eternity. We also had to wait a long time for the departure check-out and the security check-up was still closed until shortly before boarding. Arriving three hours before departure, the Tehran Imam Khomeini Airport is not really too early.

Despite some delay of our first flight from Tehran to Vienna we got our connection flight to Hamburg. Also our luggage we received without problems in Hamburg. Tired but satisfied with the work done at the competition, we went to the Hamburg sun on a cold Sunday morning. Most of us needed a bit of sleep.