This year we got two papers accepted for the RoboCup Symposium 2017!

The first paper, by our teammate Kenzo Lobos  is titled “Toward Real-Time Decentralized Reinforcement Learning using Finite Support Basis Functions”. Kenzo used reinforcement learning techniques to develop an in-walk kick. We are very proud of the results that proves this aproach to be superior to its classical counterparts.

The second paper was presented by Nicolas Cruz and Kenzo Lobos, both members of out team. In the paper “Using Convolutional Neural Networks in Robots with Limited Computational Resources: Detecting NAO Robots while Playing Soccer” they explain how they wer able to deploy Convolutional neural Networks on the Nao robot. The results are very good and we are very exited to use this method on our matches!

We are very proud to inform that we got 4th place in the robocup 2016 in Leipzig!
After a year of hard work, we set up to travel to Leipzig to represent Chile. We had some difficult games but we managed to come on top. We finally lost to this year champions: the Bhuman team from germany which left us in a 4th place of which we are very proud of. We want to thank all our fans for watching our games and for suporting us through all these years.
More challenges lie forward along our path to the 2017 Robocup, and we can’t wait to tackle them head on!

We are glad to announce that this year we got two papers accepted for the interactive session of the RoboCup Symposium 2016!

The first work was developed by Pablo Cano, who is one of our team captains and is pursuing a master degree in electrical engineering. Pablo applied techniques of multi-hypothesis tracking with random finite sets to improve the obstacle map that is currently used in our team. The results are amazing, and we are very excited about using it in our matches.

The second paper was presented by Leonardo Leottau, doctoral student, covering part of the research done during an internship in the Delft Center for Systems and Control of the Delft University of Technology, in the laboratory of Prof. Robert Babuska. The paper is an application of decentralized reinforcement learning in mobile robotics, showing applications in a three dimensional mountain car, as well as in both simulated and real setups with a differential robot.

Today we received an email from the SPL Technical Committee indicating that we are classified for the RoboCup 2016. We are very excited for playing against such incredible teams in Germany.
We are working hard towards keeping our position of this year, which require several (and drastic) improvements in some core modules of our code. Our main challenges, despite the realistic ball and the robot perception stated in the new rules, mainly concern our gait stability and actuation stuff, as well as some perception issues we noticed during our matches in China.

(Having passed a long time without keeping our website updated , we are working towards fixing this issue. Sorry guys)

We gladly announce that we got the fourth place in the SPL again! Furthermore, 3 of our team mates presented papers in the RoboCup Symposium (Pablo, Carlos and Matías), and Carlos won the Best Science Paper Award!

It was a very long trip from Chile to Hefei of about 60 hours (2 planes, 1 bus, 1 train and connections of up to 20 hours), but it definitely worth it. China was a great experience because it was our first time in Asia, we met amazing people and we visited awesome places.

We hope to keep doing as well as this year, we have great challenges for the RoboCup 2016.

Taken in Joao Pessoa, Brazil

The team 2014. From left to right, up to down: Leonardo Leottau, Matías Mattamala, Pablo Cano, Prof. Javier Ruiz-del-Solar (Team Leader), José Miguel Yañez (Captain), Matías Silva and Pablo Saavedra.

Sooo, after a really hard competition in Joao Pessoa, when we had a pretty bad start with a draw (RoboCanes, USA), two lost games (B-Human and Berlin United, both from Germany) and a dramatic win in the round robin (Philosophers, Estonia), we could pass to the intermediate round. There we defeated UPennalizers (USA), reaching the quarterfinals to playing against NaoDevils (Germany), to whom we won by a penalty shootout. Then, in semifinals, we played against Nao Team HTWK (Germany), falling by 4:5 (you must see the video below). It gave us a last chance to reach the third place, but we also lost against a known partner, B-Human. You can see a full tracking of the matches in the official RoboCup SPL Results site.

However, despite losing our last two games, we’re really excited about our performance this year. We had a lot of problems, because our robots were demanded more than ever (for this reason they started to feel a bit “tired”, falling down repeatedly during the games), but we could cope this issues and we got nice results. Now we already returned to Chile and we started working for the next RoboCup, which will take place in Hefei, China!

We’ll be announcing any news, but for now is time to go back to work!