Robot Teams

Carologistics

Institute: RWTH Aachen University and Aachen University of Applied Sciences
Website: http://www.carologistics.org/

The IMA/ZLW & IFU Institute Cluster, RWTH Aachen University, the Knowledge-based Systems Group, RWTH Aachen University, and the Department for Electrical Engineering and Information Technology, Robotics Group, FH Aachen founded the Carologistics robot team in 2012 to participate in the Festo RoboCup Logistics League

Starting out with a Hackathon, a young team of researchers and students was formed to participate in the RoboCup 2012 competition in Mexico City. Within the very short time of just three weeks this team was able to create a system that could tackle the given task of feeding a managing a (simplified) processing chain of machines. This was especially possible because the team could built its applications on top of the already existing Fawkes framework providing parts of the basic functionality already. As part of the project, code that implements the Adaptive Monte-Carlo Localization (AMCL) was ported from ROS into a Fawkes plugin. Additionally, through the tight integration with ROS, the ROS locomotion part of the navigation stack could be used for autonomous driving including collision avoidance. Both, laser-based self-localization with a map and navigation including collision avoidance was a first for the new Logistics League sponsored by Festo based on the Robotino robots. Further the team employed an omni-directional vision system, allowing it to spot pucks all around the robot base without turning, again a novelty for the competition.

Team members of the Carologistics team now participate in both, the Technical Committee as well as the Organization Committee.

Robots

Feed

Campus52 TV snippet about the Carologistics

Posted on July 18, 2014 at 12:28 PM by Carologistics RoboCup Team RSS

The Campus52 campus TV has created a short documentary about our preparations for the RoboCup 2014 competitions in Brazil. We think it's great and thank them for their efforts!

We have now arrived and are preparing in the hotel in Joao Pessoa. You can find a flow of posts about our progress on Twitter with the #Carologistics hashtag.

Read this post in context »

Winner RoboCup German Open 2014!

Posted on July 21, 2014 at 1:53 PM by Carologistics RoboCup Team RSS

We won! The final ended 71:30, the team is celebrating! Even the professor is pleased...



Read this post in context »

Playoffs: Three games, three victories!

Posted on July 21, 2014 at 1:53 PM by Carologistics RoboCup Team RSS

By performing a perfect playoff round we entered the final for the first time in the teams existence! We scored 54:23, 76:47, and 56:18 (The righthand pic is not completely up to date, we passed the 500 points :-)). This means we'll face Munich in the final. Their robots still face some issues with lacking collision avoidance, which resulted in their robots being taken out in earlier games. We are carefully optimistic that we'll deliver a good final game, however we are still very surprised of our success...
All in all we are very happy with the outcome so far, there is still lots to do, right now we try to fix some minor bugs, but only very carefully, of course. Maybe it's a good idea to get some sleep this night to reduce the risk of human casualties tomorrow :-) Our final start around 10:30am, so please keep your fingers crossed!


Read this post in context »

Qualification Phase: So far, so good!

Posted on July 21, 2014 at 1:53 PM by Carologistics RoboCup Team RSS

Our final game of the qualification phase was the best one so far, we scored 81 Points! This puts us well in front of the other teams. As hoped vision didn't cause us any problems, and our agent strategy with different roles for each robot turned out well. There is still some debugging to do, but we do have some time now until the play offs start.
During play offs each team will directly compete with an opposing team, the team that scores more points in game will be awarded 3 points. This is potentially very dangerous for us since the number of points scored in the game is not taken into account. One total slip, due to hardware failures or similar, could cost us the finals regardless of our overall performance. So please keep your fingers crossed!!

Read this post in context »

Day 1 of the RoboCup GermanOpen has ended!

Posted on July 21, 2014 at 1:53 PM by Carologistics RoboCup Team RSS

The first day of the tournament has ended, and with it the bigger part of the qualification phase for the play offs. Currently we are leading the highscore list! In each game we had all our three robotinos on the field (for most of the time) and could score max points in the exploration phase.


Our agent strategy paid off when one of the robots had to be taken out and the remaining robots automatically took over it's remaining tasks. While the exploration phase was a reliable point warrant we could not get much out of the production phase. Reason for that was an unexpected effect: We had trained our color model for the vision based signal detection on single signal lights. However, in the production phase, all signals switch to green. This gave the playfield a slightly green ambience light, which our color model couldn't cope with. As a result our robots placed pucks under machines, could not detect the successful production of a good and subsequently left to try again. Nevertheless we could some points in some seldom cases where the vision managed to detect the signal change. As a result we could score at least 58 points in each game and even 74 points in our best run! This put us well on front of the others! And regarding the vision problems we have good news: We fixed it!


We will use different color models for the different phases of the game to adapt to the different light conditions on the field. Since we are already safe for the play off phase, we will use our remaining game tomorrow morning for excessive testing. In the following play offs the game will become a little bit more rough: The two teams playing on the same field will have to cross ways more often which will surely lead to lots of collisions. Since we have implemented reliable collision avoidance from on the beginning we deem ourselves to be well prepared... We'll find out :-)

Read this post in context »