Content. 3 Preface 4 Who We Are 6 The RoboCup Initiative 7 Our Robots 8 Hardware 10 Software 12 Public Appearances 14 Achievements 15 Interested?

Similar documents
FU-Fighters. The Soccer Robots of Freie Universität Berlin. Why RoboCup? What is RoboCup?

FalconBots RoboCup Humanoid Kid -Size 2014 Team Description Paper. Minero, V., Juárez, J.C., Arenas, D. U., Quiroz, J., Flores, J.A.

RoboCup. Presented by Shane Murphy April 24, 2003

Baset Adult-Size 2016 Team Description Paper

FUmanoid Team Description Paper 2010

EROS TEAM. Team Description for Humanoid Kidsize League of Robocup2013

Hierarchical Controller for Robotic Soccer

IRH 2017 / Group 10. Hosen Gakuen High School Risu inter. Takeru Saito, Akitaka Fujii. Theme3 Most advanced technologies of robots

KMUTT Kickers: Team Description Paper

NimbRo 2005 Team Description

ZJUDancer Team Description Paper Humanoid Kid-Size League of Robocup 2014

ZJUDancer Team Description Paper Humanoid Kid-Size League of Robocup 2015

Robo-Erectus Jr-2013 KidSize Team Description Paper.

How Students Teach Robots to Think The Example of the Vienna Cubes a Robot Soccer Team

Intelligent Humanoid Robot

ZJUDancer Team Description Paper

Humanoid robot. Honda's ASIMO, an example of a humanoid robot

Kid-Size Humanoid Soccer Robot Design by TKU Team

Nao Devils Dortmund. Team Description for RoboCup Matthias Hofmann, Ingmar Schwarz, and Oliver Urbann

Team Description Paper: HuroEvolution Humanoid Robot for Robocup 2014 Humanoid League

Task Allocation: Role Assignment. Dr. Daisy Tang

KUDOS Team Description Paper for Humanoid Kidsize League of RoboCup 2016

CORC 3303 Exploring Robotics. Why Teams?

Construction of Mobile Robots

YRA Team Description 2011

BehRobot Humanoid Adult Size Team

A Lego-Based Soccer-Playing Robot Competition For Teaching Design

Robocup Electrical Team 2006 Description Paper

Team Description Paper: HuroEvolution Humanoid Robot for Robocup 2010 Humanoid League

RoboCup TDP Team ZSTT

WF Wolves & Taura Bots Humanoid Kid Size Team Description for RoboCup 2016

Extra Curricula. Robotics

Courses on Robotics by Guest Lecturing at Balkan Countries

Team KMUTT: Team Description Paper

Team Description Paper: Darmstadt Dribblers & Hajime Team (KidSize) and Darmstadt Dribblers (TeenSize)

Hanuman KMUTT: Team Description Paper

Does JoiTech Messi dream of RoboCup Goal?

RoboCup: Not Only a Robotics Soccer Game but also a New Market Created for Future

SitiK KIT. Team Description for the Humanoid KidSize League of RoboCup 2010

Team Description 2006 for Team RO-PE A

Team MU-L8 Humanoid League TeenSize Team Description Paper 2014

EDUCATIONAL ROBOTICS' INTRODUCTORY COURSE

Dutch Nao Team. Team Description for Robocup Eindhoven, The Netherlands November 8, 2012

The description of team KIKS

Alternative English 1010 Major Assignment with Activities and Handouts. Portraits

RoboCup was created in 1996 by a group of Japanese,

CIT Brains (Kid Size League)

Overview Agents, environments, typical components

GermanTeam The German National RoboCup Team

Development and Evaluation of a Centaur Robot

MRL Team Description Paper for Humanoid KidSize League of RoboCup 2017

Multi-Platform Soccer Robot Development System

Learning and Using Models of Kicking Motions for Legged Robots

Berlin United - NaoTH 2014

Application from Hamburg Bit-Bots for RoboCup 2017

RoboPatriots: George Mason University 2010 RoboCup Team

Team RoBIU. Team Description for Humanoid KidSize League of RoboCup 2014

Robotic Systems ECE 401RB Fall 2007

Franco German press release. following the interview between Ministers Le Maire and Altmaier, 18 December.

1 Abstract and Motivation

Representation Learning for Mobile Robots in Dynamic Environments

Hierarchical Case-Based Reasoning Behavior Control for Humanoid Robot

The future of research on safety and security in Germany - Results from an explorative Delphi study

Nao Devils Dortmund. Team Description for RoboCup Stefan Czarnetzki, Gregor Jochmann, and Sören Kerner

Nao Devils Dortmund. Team Description for RoboCup 2013

Humanoid Robots. by Julie Chambon

Keywords: Multi-robot adversarial environments, real-time autonomous robots

ROBOTSOCCER. Peter Kopacek

Eleonora Escalante, MBA - MEng Strategic Corporate Advisory Services Creating Corporate Integral Value (CIV)

RoboCup 2013 Humanoid Kidsize League Winner

Robot: icub This humanoid helps us study the brain

Concept and Architecture of a Centaur Robot

Learning and Using Models of Kicking Motions for Legged Robots

2 Our Hardware Architecture

Team Description for Humanoid KidSize League of RoboCup Stephen McGill, Seung Joon Yi, Yida Zhang, Aditya Sreekumar, and Professor Dan Lee

The Attempto Tübingen Robot Soccer Team 2006

Humanoid Robot NAO: Developing Behaviors for Football Humanoid Robots

COMP219: Artificial Intelligence. Lecture 2: AI Problems and Applications

Children and Social Robots: An integrative framework

Concept and Architecture of a Centaur Robot

Sensor system of a small biped entertainment robot

Application from Hamburg Bit-Bots for RoboCup 2015

Optic Flow Based Skill Learning for A Humanoid to Trap, Approach to, and Pass a Ball

Artificial Intelligence Paper Presentation

STOx s 2014 Extended Team Description Paper

Self-Localization Based on Monocular Vision for Humanoid Robot

MRL Team Description Paper for Humanoid KidSize League of RoboCup 2013

RoboPatriots: George Mason University 2014 RoboCup Team

Autonomous Systems at Gelsenkirchen

CITBrains (Kid Size League)

TU Graz Robotics Challenge 2017

Robo-Erectus Tr-2010 TeenSize Team Description Paper.

SPQR RoboCup 2016 Standard Platform League Qualification Report

Growing up with Robots Costa MFM and Fernandes JF

Birth of An Intelligent Humanoid Robot in Singapore

Making Representations: From Sensation to Perception

US Darts Masters Training Guide

Using Reactive and Adaptive Behaviors to Play Soccer

Cooperative Behavior Acquisition in A Multiple Mobile Robot Environment by Co-evolution

CS343 Introduction to Artificial Intelligence Spring 2012

Transcription:

Content 3 Preface 4 Who We Are 6 The RoboCup Initiative 7 Our Robots 8 Hardware 10 Software 12 Public Appearances 14 Achievements 15 Interested? 2

Preface Dear reader, Robots are in everyone's minds nowadays. In science fiction movies and books they are shown to be powerful machines, capable of extraordinary deeds. Although we're not quite there yet, researchers are working to realize the dream of intelligent machines that will assist us in our lives and make the world a better place. Robots are an inherent part of the technical evolution we will experience in the near future. In RoboCup, where teams of autonomous robots play soccer against each other, universities all over the world are trying to tackle the core problems at the heart of robotic research. The game of soccer poses an extreme challenge for today's robots: stable walking, perceiving the world visually, interacting with a highly dynamic environment and cooperating with other robots are all problems that have yet to be fully solved. Solutions for these problems are not only necessary to play soccer, but are also essential in many other areas: e.g. household aids for senior citizens, rescue operations in dangerous environments or autonomous vehicles. With its research, our team is not only making an active contribution to the creation of current and future robot generations, but is helping to solve societal challenges, too. The FUmanoids team 3

Who We Are The FUmanoids are a student-based project at the Freie Universität Berlin, led by professor Dr. Raúl Rojas. Their name is an amalgamation of Freie Universität (FU) and the English word, humanoids. Founded in 2006, the team of bachelor and master students develops humanoid robots and works in the fields of artificial intelligence, robotics, computer vision, electronics and mechanics. The aim of the project is to develop autonomous, soccer-playing robots. Any work the team members spend on the robots is done voluntarily, as the project is not part of the university curriculum. However, the university supports the team by providing equipment and a work space. The project is a unique opportunity to apply what has been learned and to pursue personal research interests. Creative ideas can be tested and verified on actual robots. Additionally, students gather valuable experience while working with complex software in a professional team. The team consists of six humanoid robots. On average, the robots were designed and programmed by a dozen bachelor and master students. 4

5

The RoboCup Initiative RoboCup is an international initiative to help facilitate the exchange and comparison of knowledge in the field of robotics. It was founded with the goal to beat the human soccer world champions in a fair match by the middle of the 21st century. This new standard challenge within artificial intelligence (AI) research allows us to compare and experience any progress made in the robotics field. It increases the exchange between scientists from all over the world. Competitions Teams from all around the globe compete in national competitions (e.g. RoboCup German Open) and the annual RoboCup world championship. These events cover multiple robot leagues. More than 2,600 participants from 45 countries participated in the 2013 RoboCup competition in Eindhoven. The championship attracted over 40,000 visitors. RoboCup events are covered by both national and international media outlets. Humanoid League The FUmanoids participate in the Humanoid League, one of the most popular RoboCup soccer leagues. They play in the kid size category where 24 teams have their robots play in 4 vs 4 matches. These robots are up to 90cm in height and are designed, built and programmed by the teams themselves. 6

Our Robots The FUmanoid team currently consists of six robots. The newest model is an improved version of our robots from 2012. The robots use a web cam to gather visual information about their surroundings. Using three gyroscopes and acceleration sensors they maintain their balance while walking on two legs. They play autonomously and select where to walk and when to kick the ball. If they fall down, they get right back up again and continue playing. Design: Height: 65cm Weight: 4,5kg Degrees of Freedom: 20 Hardware: CPU: ODROID-X2 mit 4 x 1.7 GHz Webcam: Logitech c920 ( 640 480 ) OS: Linaro (based on Ubuntu) 7

Hardware The rules of the Humanoid League dictate many aspects of the robots design. They are updated every year to facilitate the development of new designs and to get closer to the 2050 goal of winning a fair match against the FIFA champions. The rules require the robots to roughly have human proportions. They are only allowed to use sensors that resemble humanlike senses, e.g. vision, touch or sense of balance. Contrary to human soccer players, they may use WiFi to communicate with each other and with the referee. Fulfilling all soccer-related tasks requires a coherent hardware concept. Additionally, the team has to consider limitations of weight, battery capacity and computational power in their design. Finally, funding sometimes has a limiting factor in how far a desired hardware design can be realized. 8

Power Board The power board is a smart power-supply circuit board the team developed in early 2013. It has undergone a process of constant improvement ever since. The board not only provides the whole robot with the correct current, but also detects errors in the electronic system. If a short circuit occurs, the power board protects the electronic components by switching them off in less than two milliseconds. Additionally, the current battery charge level can easily be seen by the human handlers thanks to a colored LED. Stereo Vision Currently, almost all teams in the Humanoid League use a single camera to gather visual data. This approach comes with limitations, as no depth perception is available. Therefore the FUmanoids have started integrating a stereo camera, developed by our research group at Freie Universität. This camera operates in a similar way to the human eyes. It provides the robots with accurate depth information which allows them to identify the distance and location of objects more precisely. 9

Software The robots are only able to work efficiently with powerful software. It prepares and processes sensor data, controls the hardware and makes decisions about the robots' behavior. The FUmanoids show a high degree of creativity and innovation in this area. Their success in competitions is based on the development and implementation of new technology. Currently the focus lies on three main areas: Image Processing Image processing extracts detailed information from the camera images. With the help of color-classified images and object extractors, the position of the two goals, the field lines, the ball and other robots are determined. Based on this data, the behavior control makes decisions. The software has to overcome difficulties like blurred images as a result of walking, changing lighting conditions or, thanks to the colorfully dressed audience, a highly dynamic background. 10

Position Data After the image-processing has determined the objects on the playing field and calculated their coordinates relative to the robot, the software then creates a global world model based on this data. This allows the player to guess the position of objects very precisely, even if they cannot be seen at that precise moment. Thus, the position of the ball rolling behind another robot can be predicted based on the physical properties of its movement - although the ball is currently hidden. Localization Based on the world model and visual data, the robots will constantly try to locate themselves on the playing field. Knowing where the robot itself and all other objects are, is an important prerequisite for effectively planning the next moves. A particular challenge is the symmetry problem. Both halves of the playing field are absolutely identical and own goals have already been the downfall of many teams. Our current approach combines several types of information to aid this process. On the one hand, the robot uses the position of non-moving objects like goal posts. On the other hand, the positions of team members can provide orientation. In addition, the goalie acts as a corrective function for other players estimations, as it is in a well defined and easy-to-verify position throughout the game. 11

Pu blic Appearan c es The FUmanoids have demonstrated their skills at various events. They have participated in tournaments ranging from 2,000 visitors at local competitions to 40,000 visitors at the RoboCup world championship. Aside from students and people interested in technology, families with children make up the bulk of the audience. The FUmanoids are visited regularly by camera teams from national and international TV stations and are interviewed by print and online media. In doing so, the robots have already reached a broad audience. Additionally, we also publish updates on our homepage, our YouTube channel or on Facebook for those who are interested in our research and progress. Event Visitors Participants Robocup 40,000 2,456 German Open 3,500 950 Iran Open 5,000 1,000 Long Night of Science 2,000 2 Teams 12

13

Achievements 1st place RoboCup German Open 2014 1st place RoboCup Iran Open 2014 4th place RoboCup German Open 2013 2nd place RoboCup German Open 2012 4th place Robocup 2011 (Turkey) 2nd place RoboCup 2010 (Singapore) 2nd place RoboCup 2009 (Austria) 2nd place RoboCup German Open 2008 3rd place RoboCup 2007 (USA) At the 2012 and 2013 RoboCup competitions the FUmanoids reached the quarterfinals. The team did not participate in RoboCup 2014. 14

Interested? Got interested? On the back of this brochure you will find our contact details. We are happy to answer all of your questions. Are you a student? The FUmanoids offer topics for bachelor and master theses as well as software projects on a regular basis. We are always happy to welcome new team members. Support us! The development of modern autonomous robots is a costly undertaking. As a student-based project we are grateful for every sponsor and are happy to offer individual sponsoring deals. Don't hesitate to contact us! 15

Version: February 2015 AG Intelligente Systeme und Robotik Freie Universität Berlin Institut für Informatik Arnimallee 7 14195 Berlin Germany info@fumanoids.de https://www.facebook.com/fumanoids http://www.fumanoids.de Photos: FUmanoids, Kristina Schippling, Thomas Trutschel, Jakob Weber