RoboCup@Home aims to develop service robots for personal domestic use through competitions evaluating robots' capabilities in realistic home environments. The document discusses service robotics research and development (R&D) focusing on an educational robotics initiative using open robot platforms to promote robotics education through competitions and outreach programs.
Report
Share
Report
Share
1 of 88
Download to read offline
More Related Content
RoboCup@HomeEDU AI-Focused Robotics Education by Home Service Robot DIY | Victoria October 15, 2019
1. Service Robotics R&D
RoboCup@Home EDUCATION
AI-Focused Robotics Education by Home Service Robot DIY
Victoria University 2019-10-15 | Jeffrey Too Chuan TAN
2. Jeffrey Too Chuan TAN(陈图川)
[ Education Background ]
2007 - 2010 The University of Tokyo (Japan), Department of Precision Engineering, Doctor of Engineering
2004 - 2007 Universiti Tenaga Nasional (Malaysia), Master of Mechanical Engineering
1999 - 2003 Universiti Tenaga Nasional (Malaysia), Bachelor of Mechanical Engineering (Hons.)
[ Working Experience ]
2017 - Present Associate Professor, Nankai University (China)《天津市青年千人计划》
2017 - Present Research Fellow, Tamagawa University (Japan)
2014 - 2017 Project Assistant Professor, Institute of Industrial Science, The University of Tokyo (Japan)
2015 - 2017 Adjunct Lecturer, Tokyo City University (Japan)
2013 - 2014 Project Researcher, Institute of Industrial Science, The University of Tokyo (Japan)
2011 - 2013 Project Researcher, National Institute of Informatics (Japan)
2010 - 2011 Project Researcher, Graduate School of Engineering, The University of Tokyo (Japan)
2004 - 2007 Tutor, Universiti Tenaga Nasional (Malaysia)
[ Professional Services ]
2016 - Present Committee (Service and Junior), World Robot Summit
2016 - Present Organizing Committee, RoboCup Federation (@Home)
2015 - Present Committee, RoboCup@Home Education
2014 - Present Organizing Committee, RoboCup Japan (@Home)
2
Profile
3. RoboCup@Home
RoboCup@Home aims to foster the development of service and assistive robot technology to make possible
future personal domestic applications. The competitions comprise of a set of benchmark tests to evaluate the
robots’ capabilities in realistic home environment settings and scenarios, with the research focuses on: human-
robot interaction and cooperation, navigation in dynamic environments, computer vision and object recognition
under natural light conditions, object manipulation, adaptive behaviors and learning, ambient intelligence, and
system integration.
3
4. Outline
1. Service Robotics R&D
2. Prologue: Team KameRider
3. RoboCup@Home EDUCATION Initiative
a. Education Challenge
b. Educational Open Robot Platforms
c. Outreach Programs
6. Home Service Robot DIY
a. Robot LEG
– Autonomous Navigation
b. Robot EYES
– Visual Perception
c. Robot ARM
– Object Manipulation
d. Robot MOUTH
– Human-Robot Interaction
e. Robot BRAIN
– AI, Machine Learning,
Cloud Computing, Big Data
6
8. Robot EYES – Visual Perception
• Image Processing by OpenCV
• Deep Learning Object Detection by YOLO
8
9. Robot EYES – Visual Perception
• Person recognition result in RoboCup 2016
9
10. Improving Deep Learning Based Object Detection by
CycleGAN Method Under Inconsistent Illumination Conditions
10
Three illumination conditions of
the real environment
CycleGAN is used to realize the mutual
transformation of scenes
Dark environment before
brightness enhancement
Dark environment after
brightness enhancement
Object detection after
brightness enhancement
The top view of the
visual task scene and
the robot vision with
supplementary light
Object detection
confidence level
improvement
11. Robot LEG – Autonomous Navigation
11
• Indoor Autonomous Navigation
– Adaptive Monte Carlo Localization (AMCL)
– Simultaneous Localization and Mapping (SLAM)
– Static and Dynamic Obstacle Avoidance
12. Robot ARM – Object Manipulation
12
http://wiki.ros.org/turtlebot_block_manipulation
Object Manipulation
13. Multi-Object Grasp Planning in High Distribution Density
using Inverse Reachability Map and Base Repositioning
13
Experiment environment and object distribution IRM of different type of objects
System components and operation flow
Experiment results
15. Human-Robot Interaction (HRI)
HRI Research, Development and Applications
Human-Robot Collaborative Work
15
Human-robot collaborative cell production system
17. Robot BRAIN – AI, Machine Learning, Cloud Computing, Big Data
17
Client Systems
Robot Learning
Knowledge
Transfer
Cloud System
• Processing Servers
• Databases
?
18. Robot BRAIN – AI, Machine Learning, Cloud Computing, Big Data
• Crowdsourcing of virtual HRI for collaborative strategy learning
18[J. T. C. Tan, Y. Hagiwara, T. Inamura, “Robot Learning Framework via Crowdsourcing of Human-Robot Interaction
for Collaborative Strategy Learning,” in Proc. of the 24th IEEE RO-MAN (Interactive Session), IS04, 2015]
19. State parameters:
• Self
• Action
• Object(Target)
• Location
𝑆𝑒𝑙𝑓_𝐴𝑐𝑡𝑖𝑜𝑛𝑖 = 𝑓 𝑆𝑒𝑙𝑓_𝐴𝑐𝑡𝑖𝑜𝑛𝑖−1, 𝑃𝑎𝑟𝑡𝑛𝑒𝑟_𝐴𝑐𝑡𝑖𝑜𝑛𝑖, 𝑊𝑜𝑟𝑘_𝐶𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑖
𝐴𝑔𝑒𝑛𝑡_𝐴𝑐𝑡𝑖𝑜𝑛(𝑂𝑏𝑗𝑒𝑐𝑡, 𝐿𝑜𝑐𝑎𝑡𝑖𝑜𝑛)
𝑊𝑜𝑟𝑘_𝐶𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛 = 𝑂𝑏𝑗𝑒𝑐𝑡1(𝐿𝑜𝑐𝑎𝑡𝑖𝑜𝑛), … , 𝑂𝑏𝑗𝑒𝑐𝑡 𝑛(𝐿𝑜𝑐𝑎𝑡𝑖𝑜𝑛)
“Minimum information” to describe the current state
Collaborative Intelligence
19
• Partner
• Action
• Object(Target)
• Location
• Work
• Action(Static)
• Object1-n
• Location1-n
• Condition1-n(Omitted)
20. Extraction of Embodied Collaborative
Behaviors from Cyber-Physical HRI with
Immersive User Interfaces
• Contents
– (See) Visual Observation
• Movement of HMD to
determine observed target
– (Say) Verbal Communication
• Spoken speech
– (Do) Action
• Agent’s body movement to
determine traveled path
• Timing
– Contents’ occurrence timings
w.r.t. collaboration operation
22. Robot BRAIN – AI, Machine Learning, Cloud Computing, Big Data
• Crowdsourcing of virtual HRI for collaborative strategy learning
22
23. Robot BRAIN – AI, Machine Learning, Cloud Computing, Big Data
• Crowdsourcing of virtual HRI for collaborative strategy learning
23
Handyman (GPSR) Interactive Clean Up
Human Navigation
31. 2013 The Beginning of Team KameRider
2013.05.03-06 RoboCup Japan Open 2013 Tokyo, Japan
• [UT] Jeffrey
• [Award] JSAI Award [SIGVerse for RoboCup @Home Simulation]
• [Award] RoboCup @Home Simulation [2nd Place]
2013.06.24-07.01 RoboCup 2013 Eindhoven, Netherlands
(International)
• [Symposium] Poster: “Open Web Based Development Platform for
RoboCup @Home Simulation”
• [Symposium] Oral: “Development of RoboCup@Home Simulation
towards Long-term Large Scale HRI”
32. 2014 Forming Collaboration
UT-NKU (China), UT-UTM (Malaysia)
2014.03-06 Internship of Mr. Tey @ SIT, Japan
• [Internship] Mr. Tey (UTM) assisted Jeffrey's team in the
development of a basic robot platform for RoboCup
@Home
2014.05.03-06 RoboCup Japan Open 2014 Fukuoka, Japan
• [UT] Jeffrey, [NKU] 6 members, [UTM] Tey Wei Kang
• [Award] JSAI Award [Standard Platform for RoboCup
@Home]
• [Award] RoboCup @Home Simulation [2nd Place]
33. 2014 Forming Collaboration
UT-NKU (China), UT-UTM (Malaysia)
2014.06-09 Internship of Mr. Seow @ UT, Japan
• [Internship] Mr. Seow (UTM) develops the basic robot
platform for RoboCup @Home based on the RCF support
2014.12.06 Intelligent Home Robotics Challenge 2014, Tokyo
• [UT] Jeffrey, [UTM] Lim Kian Sheng, Mohamad Hafizuddin
bin Majek, Muhammad Faiz bin Muhammad Rozi
• [Award] Mobile Robot Category 3rd Place
• [Award] Overall 3rd Place
34. 2015 Starting Education Challenge
2015.05.03-06 RoboCup Japan Open 2015 Fukui, Japan
• [UT] Jeffrey, [NKU] 3 members, [UTM] Muhammad
Najib Abdullah, Nicole Tham Lei May
• [Award] RoboCup @Home SPL (Beta) [1st Place]
• [Award] RoboCup @Home Simulation [3rd Place]
35. 2015 Entering International RoboCup
2015.07.17-23 RoboCup 2015 Hefei, China
(International)
• [UT] Jeffrey, [NKU] 7 members, [UTM]
Yeong Che Fai, Seow Yip Loon, Nicole Tham
Lei May
• Overall ranked 7th out of 17 qualified teams
• Top 9 teams to enter Stage 2
36. 2016 Collaborative Team UT-NKU-UTM-SIT
2016.03.24-27 RoboCup Japan Open 2016 Aichi,
Japan
• [Award] RoboCup @Home Education [2nd Place]
• [Award] RoboCup @Home Simulation [1st Place]
2016.06.30-07.04 RoboCup 2015 Leipzig, Germany
(International)
• Overall ranked 7th out of 23 qualified teams
36
37. 2017 Collaborative Team NKU-UTM-SIT
RoboCup Japan Open 2017 Nagoya
• [Award] RoboCup @Home Education [1st Place]
• [Award] RoboCup @Home Simulation [2nd Place]
RoboCup 2017 Nagoya (International)
• [Award] RoboCup @Home SSPL [Overall ranked 4th
out of 7 qualified teams]
RoboCup Asia-Pacific 2017 Bangkok
• [Award] RoboCup @Home [1st Place]
• [Award] RoboCup @Home Education [1st Place]
37
41. AI-Focused Robotics Education by
Home Service Robot DIY
The “Bridging Problem”
School-level Robotics Education vs University-level Robotics Research
• Bottom-up vs Top-down
• Conceptual Problems vs Real World Problems
The Blooming of AI, Cloud and Big Data
• Learning Platform and Ecosystem
41
42. RoboCup@Home EDUCATION
RoboCup@Home EDUCATION is an educational initiative that
promotes educational efforts to boost RoboCup@Home
participation and service robot development.
Under this initiative, currently there are 3 projects in
operation:
1. RoboCup@Home Education Challenge
2. Support the Development of Educational Open Robot
Platforms for RoboCup@Home (service robotics)
3. Outreach Programs (domestic workshops, international
academic exchange programs, etc.)
http://www.robocupathomeedu.org/
https://www.facebook.com/robocupathomeedu/
42
46. RoboCup@Home Education Challenge
• RoboCup@Home (Main)
– Since 2006
• RoboCup@Home Education Challenge
– RoboCup Japan Open 2015, Fukui (SPL Beta), Japan
– RoboCup Japan Open 2016, Aichi, Japan
– RoboCup Japan Open 2017, Nagoya, Japan
– RoboCup Asia-Pacific 2017 Bangkok, Thailand
– RoboCup Japan Open 2018, Ogaki, Japan
– European RoboCupJunior Championship (EURCJ) 2018, Montesilvano,Italy
– RoboCup 2018 Montreal, Italy
– RoboCup China Open 2019, Shaoxing, China
– European RoboCupJunior Championship (EURCJ) 2019, Trieste, Italy
– RoboCup 2019 Sydney, Australia
– RoboCup Japan Open 2019 Nagaoka, Japan (August)
• Upcoming events
– RoboCup Junior Australia Open 2019 Melbourne, Australia (October)
– RoboCup Asia-Pacific 2019 Moscow, Russia (November)
– RoboCup Japan Open 2020 Nagoya, Japan (March)
– RoboCup 2020 Bordeaux, France (July) 46
48. RoboCup@Home Education
RoboCup Japan Open 2015, Fukui (SPL Beta)
• Date: 2015 May 1 (Fri) - 4 (Mon)
• Participated Teams
1. AHP-1 eR@sers (Tamagawa University)
2. AHP-2 OIT Kitayama (Osaka Institute of Technology)
3. AHP-3 KameRider (The University of Tokyo, Nankai
University (China), Universiti Teknologi Malaysia
(Malaysia))
4. AHP-4 SOBITS (Soka University)
5. AHP-5 D.K.T. IcARus (Kanagawa Institute of
Technology)
6. AHP-6 TanichuCluster (Ritsumeikan University)
Ranking
No.
Team
Basic
Functionalities
Restaurant
Sub-Total(5/2)
FollowMe
Sub-Total(5/3)
5/2+5/3
Normalization
Technical
Challenge
(InternalJudges)
Total
1st AHP-3 KameRider 400 750 1150 300 300 1450 100 40 46 38 91.33
2nd AHP-6 TanichuCluster 150 250 400 50 50 450 31 33 41 40 53.52
3rd AHP-1 eR@sers 150 0 150 560 560 710 49 18 35 16 47.48
4th AHP-2 OIT Kitayama 400 0 400 250 250 650 45 19 30 12 42.75
5th AHP-5 D.K.T. IcARus 0 0 0 181 181 181 12 23 26 34 33.91
6th AHP-4 SOBITS 0 0 0 221 221 221 15 15 31 20 29.62
49. RoboCup@Home Education
RoboCup Japan Open 2015, Fukui (SPL Beta)
The RoboCup@Home rulebook of 2014 is based and 4 tests are selected as follows:
1. Basic Functionalities
• The description in section 5.2 Basic Functionalities (pg. 40-42) is referred.
• In section 5.2.1, 1. Pick and Place (pg. 40), the objects for the robot to pick up will
be located within the reach of the working envelope of the robot arm.
2. Follow Me
• The description in section 5.3 Follow Me (pg. 43-47) is referred.
• No change is made on the rules.
3. Restaurant
• The description in section 6.3 Restaurant (pg. 64-66) is referred.
• In section 6.3.2, 1. Guide phase (pg. 64) is omitted. The object and delivery
locations will be informed before the game.
• In section 6.3.2, 2. Navigation and manipulation phase (pg. 64), the objects for the
robot to retrieve will be located within the reach of the working envelope of the
robot arm.
4. Open Challenge
• The description in section 5.5 Open Challenge (pg. 52-54) is referred.
• No change is made on the rules.
[http://www.robocupathomeedu.org/challenges/robocup-home-education-challenge-2015/rules-2015]
50. RoboCup@Home Education
RoboCup Japan Open 2016, Aichi
• Date:
– Competition days: 2016 March 25 (Fri) - 27 (Sun)
– Team setup: 2016 March 24 (Thu)
• Venue:
– Aichi Institute of Technology, Aichi, Japan
• Participating Teams:
1. eR@sers (Tamagawa University)
2. OIT Kitayama (Osaka Institute of Technology)
3. KameRider (The University of Tokyo, Nankai
University (China), Universiti Teknologi
Malaysia (Malaysia), Shibaura Institute of
Technology)
4. SOBITS (Soka University)
5. WinKIT@DKT (Kanagawa Institute of
Technology)
6. TanichuCluster (Ritsumeikan University)
7. MMR (Meijo University)
8. ODENS (Osaka Electro-Communication
University)
9. Eruca (Tokyo City University)
51. RoboCup@Home Education
RoboCup Japan Open 2016, Aichi
The RoboCup@Home rulebook of 2015 is based and 4 tests are selected as follows:
1. Navigation Test
• The description in section 5.3 Navigation Test (pg. 50-53) is referred.
• No change is made on the rules.
2. Speech Recognition & Audio Detection Test
• The description in section 5.6 Speech Recognition & Audio Detection Test (pg. 59-
61) is referred.
• No change is made on the rules.
3. Restaurant
• The description in section 6.3 Restaurant (pg. 66-70) is referred.
• In section 6.3.3, 6. Delivering phase (pg. 67), the objects for the robot to retrieve
will be located within the reach of the working envelope of the robot arm (see
below).
4. Finals
• The description in chapter 7 Finals (pg. 79-80) is referred.
• No change is made on the rules.
[http://www.robocupathomeedu.org/challenges/robocup-home-education-challenge-2016/rules-2016]
65. Approach
• Open source platform
for service robot
– Startup base, cost
effective and community
support
• Current design:
– Basic robot platform
– Modular add-ons
66. Specifications
• Mobile Base
– TurtleBot2 (Kobuki)
• Perception Systems
– Kinect for Xbox 360
• Robot Arm
– TurtleBot Arm
– Elevated Platform
• User Interface
– Digital I/O
– Android interface
– Iconic robot facial expression system
• Software framework
– Navigation
– Manipulation
– Voice Interaction
– People/object recognition
68. Open Source Solution
Open robot platform for service robotics
• Open courseware
– http://www.robocupathomeedu.org/learn
– http://robotforall.org/opencourseware/
• Support wiki
– http://robotforall.org/wiki/
• Source codes
– https://github.com/robocupathomeedu/
• Demo videos
– https://www.youtube.com/user/kameriderteam
68
69. Hardware Cost
• Current hardware cost of Open Robot Platform:
69
Item Qty Cost (USD)
Mobile platform (TutleBot2) 1 1,000
Robot arm 1 600
Elevated upper platform 1 600
Motion sensor (MS Kinect) 2 500
Electronics and miscellaneous 1 300
Controller and interface system
(Laptop PC)
1 2,000
Total 5,000
PR2: ~400,000 USD
ORP: ~5,000 USD
81. International Academic
Exchange Programs
• 2017.01.09-18 SAKURA Science Program @ Japan
– Host: Tamagawa University (Japan)
– Visitor: 10 students and 1 staff from Kasetsart University
(Thailand)
• 2016.12-2017.03 RoboCup Internship @ Japan
– Host: The University of Tokyo (Japan)
– Intern: 1 student from Univerisiti Teknologi Malaysia
(Malaysia)
• 2016.02.26-03.06 SAKURA Science Program @ Japan
– Host: The University of Tokyo (Japan)
– Visitor: 10 students and 1 staff from Nankai University (China)
• 2016.02.03-19 SAKURA Science Program @ Japan
– Host: Shibaura Institute of Technology (Japan)
– Visitor: 10 students and 2 staff from Universiti Teknologi
Malaysia (Malaysia)
• 2014.12.06 Intelligent Home Robotics Challenge 2014
@ Japan
– Venue: Tokyo
– Participated the challenge and workshop by 3 students from
Univerisiti Teknologi Malaysia (Malaysia)
• 2014.06-09 RoboCup Internship @ Japan
– Host: The University of Tokyo (Japan)
– Intern: 1 student from Univerisiti Teknologi Malaysia
(Malaysia)
• 2014.03-06 Robotics Internship @ Japan
– Host: Shibaura Institute of Technology (Japan)
– Intern: 1 student from Univerisiti Teknologi Malaysia
(Malaysia)
82. Student Development PhD Scholarship at
Australian National University
Internship in Japan Internship in ItalyInternship in Italy
83. Next Step
• Worldwide Initiative
– RoboCup@Home
Education Community
(Challenge, Workshop)
– USA, Europe (Italy),
Thailand, China, Iran,
Malaysia, Singapore, etc.
83
• Collaboration with RoboCup Junior
• Collaboration with Industrial Partners
– MathWorks, NVIDIA, ROBOTIS
• Open Courseware and Open Robot (Hardware/Software)
Development
84. Bridging Robotics Education between High School and
University: An Outreach Development in Southeast Asia
Jeffrey Too Chuan Tan1, Kanjanapan Sukvichai2, Zool Hilmi Ismail3, Ban Hoe Kwan4,
Danny Wee Kiat Ng4, Hafiz Rashidi Harun5, Amy Eguchi6 and Luca Iocchi7
MOTIVATION – There is a big gap of missing advanced skill
sets between high school and university level of robotics
education due to the differences in bottom-up and top-
down learning approaches.
SOLUTION – We aim to initiate a bridging education layer
that abstracts advanced university level robotics
development into a learning platform suitable for high
school students. The students learn by building practical
robots and competing their robots with peers.
PROJECT – We are developing a set of hardware and
software solutions as the learning platform (Fig. 1), and
organizing a series of educational activities in the form of
workshop and competition (Fig. 2). The objective of this
work is to outreach and evaluate this effort in developing
countries in Southeast Asia.
Regional Collaborators
1. Nankai University, China
2. Kasetsart University, Thailand
3. Universiti Teknology Malaysia, Malaysia
4. Universiti Tunku Abdul Rahman, Malaysia
5. Universiti Putra Malaysia, Malaysia
6. Bloomfield College, USA
7. Sapienza University of Rome, Italy
Fig. 1 Affordable robot platforms TurtleBot2 and MARRtino
Fig. 2 Outreach programs including workshop and competition
activities in China, Japan, USA and Italy (clockwise from top left)
85. World Robot Summit – Junior Category
School Robot Challenge Workshop & Trial 2017
85
86. World Robot Summit – Junior Category
School Robot Challenge Workshop & Trial 2018
86
87. Take-Home Messages
1. Service Robotics R&D
“Everyone can learn AI and Robotics!”
2. Prologue: Team KameRider
“It works!”
3. RoboCup@Home EDUCATION Initiative
a. Education Challenge
“Let’s organize together at your region!”
b. Educational Open Robot Platforms
“Give everyone a robot!”
c. Outreach Programs
“Bring us to your community!”