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!