RoboCup-nao

robocup-nao@lists.robocup.org

June 2021

2 participants
2 discussions

SPQR Team announce B-Human code usage
by Vincenzo Suriani 25 Jun '21

25 Jun '21

Dear RoboCup 2021 Standard Platform League Technical and Organizing Committees, Dear RoboCup SPL teams, in accordance with the terms and conditions of the B-Human code release license, the SPQR Team announces the use of the B-Human code for the ongoing RoboCup 2021. We use some of the native B-Human modules including walking, simulation, whistle recognition, calibration, and behavior design system. Instead, we use our own code for perception, coordination, and multi-robot distributed decision making. The following parts of the SPQR Team code are original: - Visual perception - Team coordination - Data fusion among the robots - Behaviors - Motion primitives (besides get-ups) - Reinforcement Learning Integration The following parts are modifications of the original B-Human code: - Localization - Communication network We would like to thank the B-Human Team for sharing their code and acknowledge their contribution to the RoboCup SPL. We finally apologize for the delayed communication. Kind regards, Vincenzo Suriani, on behalf of the SPQR Team

1 0

R-ZWEI KICKERS announce B-Human code usage
by Adrian Müller 15 Jun '21

15 Jun '21

Dear Robocup SPL Community, in accordance to the terms and conditions of the BHuman code release license we - the team R-ZWEI KICKERS, UAS Kaiserslautern, Germany - like to announce the use of part of the B-Human 2019 code in the upcoming competitions 2021. Currently, our code base is a modified version of the B-Human Code Drop 2019. We modified: - team behaviour control and the SimRobot environment to allow for automatic test procedures. - We developed completely custom features including the teach-in of robot behaviour (using PS4-controller, mapping key presses to skills and recording them for playback), and - a data mining analysis which clusters the robot poses to identify trigger points for our teach-in cards, and - coupled the data mining classifier to a reinforcement learning (Q-learning) module that learns the most-effective combination of rule-based and teach-in cards while playing simulated games. - Additionally, we developed our own image recognition and a multi-object detection model, based on Keras/Yolo - Currently, we develop our own kick motions and dribbling skills, and we start applying - data augmentation to increase the robustness of our object detection and - expand the automatic referee to apply the passing-challenge rules, so we can use Q-learning here as well. This email is sent respecting the code usage of the B-Human team available on their github page. R-ZWEI KICKERS would like to thank B-Human for developing this great platform! You can reach us at http://smart-machines.hs-kl.de/projekte/hs-kl-robocup/ If you are interested in more technical details please mail me for a copy of our white paper. Mit freundlichen Grüßen / Best regards Prof. Adrian Müller, PSM-1, CSM http://www.hs-kl.de/~amueller Projektleitung AK Smart-Machines http://smart-machines.hs-kl.de/ tel: +49 631/3724-5329 fax: -5313 office: H243 ------------------------------------------------------------------------------------------------------------------------- Hochschule Kaiserslautern University of Applied Sciences Fachbereich I/MST Department of Computer Sciences and Microsystem Technology Amerikastr. 1 D 66482 Zweibruecken -------------------------------------------------------------------------------------------------------------------------

1 0