Transcript Research on Robot Soccer and what PSU can do

Research on
Robot Soccer
and what PSU
can do
The perceptions
Three perceptions: see, hear and
sense_body.
 All related to one system of perception.
 Representation of perceptions:

(see 1 (ball …) …)
Visual Perception
 see.

List of objects recognized:
– type,
– direction,
– distance,
– speed,
– number.

Lines, gates, boundaries, ball.
sense_body
Force detection
 Energy available.

Actions

Principal actions:
– dash,
– kick,
– turn
– say.
Variants
of
Control
structure
– Role level
Role level
: Determines the roles of each robot.
(defender, attacker and goal keeper)
– Action level
: Selects actions of each robot.
(shooting, blocking, dribbling, etc)
– Behavior level : Move and obstacle avoidance
– Execution level : Motor control
Classification of Robot Soccer
Systems

Vision-based system
– Remote brainless system
– Brain-on-board system

Robot-based system
The system can be classified using the
location of intelligence

Selection guidelines
– Developer’s interests
– Capabilities of the robots
– Computational capabilities of
– Cost
host computer and vision system
Remote-Brainless System






A type of vision-based system

Intelligent part is implemented
in the host computer.
Centralized system
Simple and inexpensive
Easy to develop the robot
No local sensors.
Fast computing time
High cost vision system and host computer
and sampling time
Easy to debug and upgrade the program
RemoteBrainless
System

Robots
The robots consist of driving mechanism, communication part, and computational part for velocity and
for processing the data received from a host computer

Host computer
All the calculations for vision data processing, strategies, position control of robots and so on, are done
in the host computer which controls robots like radio -controlled car
Brain-on-board system




A type of vision-based system

Intelligence is partially implemented
in the host computer and robots.
Intermediate level between the centralized and the distributed systems / between
the remote-brainless and the robot based systems.
Robots can use local sensors to move to the goal and to avoid the opponent.
Can decompose the system into high level (host computer)
Easy to make
and low level (robot systems).
the system in modular form
Brainonboard
system

Robots
The robots have functions such as velocity control, position control, obstacle avoidance, etc.

Host computer
The host computer processes vision data and calculates next behaviors of robots according to strategies
and sends commands to the robots using RF modem.
Robot-based system




Distributed system

Intelligent part is
implemented in the robots.
Suitable when the large number of agents exist
Complex and expensive
Need communication among robots
Robotbased
system

Robots
The robots decide their own behavior autonomously using the received vision data, own
sensor data and strategies.

Host computer
The host computer processes only vision data
can be considered as a kind of sensor.
Main PC

Serial Port
– Select the serial communication port

Home Goal
EXAMPLE
– Select the home side on the screen

Find Objects
– Check the box of which you like to find on the field

Initial Position: tell the vision system the initial position
of each object
– E.g.) for the ball
i) turn on the radio button of ‘Ball’
ii) place the mouse on the ball and press the left button
– Repeat above procedure for another object
Main PC
EXAMPLE

Select Situation
– The situation in which the game is about to start

Command
– Click ‘Ready’: the vision system starts finding the objects
on the field
– Click ‘Start’ : the vision system starts sending commands
to the robots
– Click ‘Stop’ : the vision system stops finding objects
and sending commands
Communication (Infra-red)

Infra-red Communication
130cm
0,0cm
Transmitter
qt
Y
35cm,35cm
35cm,95cm
115cm,35cm
115cm,95cm
q t ,q r : View angle
qr
150cm
Receiver
X
– Four transmitters are used to cover the whole field
EXAMPLE
Communication

Both teams share the same transmitter via a mediator
Transmitter
PC 1 (Team A)
PC 2 (Team B)
Infra-red
module

EXAMPLE
Communication Packet
0xFF
0xFF
–
–
–
–
0xFF
0x0F
V L1
VR 1
0xAA
VL 2
VR 2
Three 0xFFs: the start of a packet
0x0F (0xF0): Team A (Team B)
VLi , VRi: left and right wheel velocity of robot i
0xAA: end of velocity data of each robot
0xAA
VL3
VR 3
0xAA
Tough, practical problems to be
solved
Real-Time image processing
 Sensor fusion (sonar, touch, vision,
light, other).
 Motors, batteries.
 Sophisticated control strategies
 Radio communication
 Agent communication

Problems

Players do not know absolute locations.

Points of reference are: boundaries,
lines and gates.

Conflicts of players
Limited visibility
 Limited communication

Software

Real-Time Problem, rule-based, agent
behavior.

Dynamic planning and execution of plans
in real-time.

Cooperation and competition.

No precise information

Non-deterministic behavior:
– results of actions are uncertain.
Software potentials (cont.)

Voice comunication of players

Various players have different rules and
behaviors, different strategies and
implementations - rule-based, neural nets,
fuzzy logic, etc.

multi-agents: competition versus
collaboration.

on-line versus off-line,

individual versus group behaviors.
FIRA and the
ECE 478/ECE 479 class at PSU
Class project and in future Capstone Project
(if a company will sponsor)
 High-School project
 Portland Cyber Theatre
 Experience in Visual C++, Basic, Lisp and
Prolog programming.
 A lot of fun.
 Travel to Japan or Korea?

Does PSU team has a chance?
1. Professor Kim from KAIST in Korea
invited us to create a new league of walking
robots
 2. We have experience with walking robots
 3. We will try to propose new competition
ideas and have a leverage of knowing them
earlier.

– Besides, our students are smart…..?
Let us look to our robots…...
What we propose
Sports
commentator

Robot soccer system
– Intelligent control system
– Multi-agent system

Composition of robot soccer system
–
–
–
–
Walking
robots
Mobile robots
Host computer
Vision system
Communication module
Team One
Complex robots
 8 to 12 servos

Our walking quadrupeds and hexapods
Multi-national team
Collaborations:
ATR, Japan
Technical University
of Warsaw, Poland
Technical University
of Gliwice, Poland
KAIST, Korea
Quadruped
Basic Radio-Controlled Spider
Hexapod with Gripper
Spider with a camera
New soccer-specialized
hexapod
Hexapod’s Soccer Kicker
Team Two
Simple robots - hexapods
 2 servos

SoccerBot
Final design of a quadruped
walker
Block Diagram of the
Simple Robot System
Communication
Signal
Motor
Power
Communication
Module
Battery
Voltage
Regulator
Micro-Controller
Logic
Power
PWM Right
PWM Left
Motor
Motor Part
Motor
Driver
Motor
Driver
Motor
Sport Commentators
Bug
Virginia Woolf
Jonas
Marvin the Crazy Robot
Class Projects Winter/Spring 2002
1. Robot position, orientation and conflict
detection using top-mounted camera
 2. Ball recognition and vision interface to the
server
 3. Server and user interface
 4. Egoistic soccer-playing strategy
 5. Collaborative soccer-playing strategy
 6. Recovery from body conflicts such as leg
entanglements of robots

Class Projects Winter/Spring 2002
7. Design of robo-head sport
commentator: speech and natural
language generation.
 8. Mechanical improvements to robots
 9. Design of a special soccer player
hexapod.

Image Processing: find position, orientation
and conflicts of a walker
Filtering, histogramming, Hough transform, equations
Localization, orientation, conflicts (leg entanglements)
Robo Soccer - Why is it
so cool?
Is Robot Soccer useful?
 Is the result of Robot Soccer useful?
 Is Robot Soccer the worlds largest
playground?
 Is there money to be made with robots
to play sports?

Why are Robot Sports so cool?




Captures the essence of soccer. (That’s why
there can be a lot of people cheering on the
teams)
”Hacker-party” more than a dull academic
conference….that is, it is a lot of activity.
Interesting problem in co-ordinating 11
processes to achieve a common goal.
The width of the field, from Robotic-nerds
to Sociologists.
Why are Robot Sports so cool?

In the simulator league, you are forced to learn
many concepts:
– for example network communication, multi-threading,
agents, hardware, etc.
Testbed for AI-algorithms, sociological
theories.
 Fairly simple environment.
 A limited set of rules (also predefined), but
always close to real world problems.

What are the problems with
Robot Competitions ?
Too much work before reaching an interesting
research level
 Too much focus on low level implementation
(still)
 Still focus on competing instead of comparing
of strategies.

Research

We do research in several areas:
– Machine learning (constructive induction)
– Decision theory
– Social agents
– walker’s gaits evolution
– +more

In all of the above we plan to use robot
sports as a testbed, or applied area.
Conclusion
Robot sports areTHE new standard problem
within AI.
 1500 researchers world-wide.
 Focuses on interdisciplinary co-operative
work between the researchers as well as
co-operation between the agents.
 New experience for PSU

Current class
4 ME students
 3 CS students
 3 EE students

Using existing robots CS and
EE students develop software
CS and EE students build new
robots from kits
6 robots of two types build
and tested in July
ME students develop 2
new robot prototypes
and kits with good
documentation
ME students adopt software to
new robots and learn
programming
Current class
4 ME students
 3 CS students
 3 EE students

Using existing robots CS and
EE students develop software
CS and EE students build new
robots from kits
6 robots of two types build
and tested in July
ME students develop 2
new robot prototypes
and kits with good
documentation
ME students adopt software to
new robots and learn
programming
Let us go deeper
Technical Publications

Hiroaki Kitano, Masahiro Fujita, Stephane Zrehen , and Koji
Kageyama, "Sony Legged Robot for RoboCup Challenge",
– In Proceedings of the IEEE INTERNATIONAL CONFERENCE ON
ROBOTICS AND AUTOMATION, IEEE, 1998, pp.2605-2612

Manuela Veloso, William Uther, Masahiro Fujita, Minoru
Asada, and Hiroaki Kitano, "Playing Soccer with Legged
Robot",
– In Proceedings of the INTERNATIONAL CONFERENCE ON
INTELLIGENT ROBOTS AND SYSTEMS, 1998, pp.437-442
How to find more about
RoboCup?

Web Pages:
http://medialab.di.unipi.it/Project/Robocup
http://www.robocup.org
 http://www.dsv.su.se/~robocup
