Quadruped Robots that Play Soccer - PowerPoint PPT Presentation

Loading...

PPT – Quadruped Robots that Play Soccer PowerPoint presentation | free to view - id: c1023-ZDc1Z



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

Quadruped Robots that Play Soccer

Description:

Quadruped Robots that Play Soccer. Claude Sammut. School of ... Stereo Microphones. Speaker. Robot Programming. The Game. The Teams. Carnegie-Mellon (USA) ... – PowerPoint PPT presentation

Number of Views:124
Avg rating:3.0/5.0
Slides: 33
Provided by: claude2
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Quadruped Robots that Play Soccer


1
Quadruped Robots that Play Soccer
  • Claude Sammut
  • School of Computer Science and Engineering
  • The University of New South Wales

2
Darren Ibbotson
Son Bao Pham
Bernhard Hengst
3
What is RoboCup?
4
Four Leagues
5
Why make robots play soccer?
  • Robots must perceive environment
  • Recognise objects
  • Form a representation of the world
  • Plan and execute actions
  • Cooperate with team mates
  • Operate in a hostile environment

6
The SONY Robots
7
Robot Programming
8
The Game
9
The Teams
  • Carnegie-Mellon (USA)
  • Essex (UK)
  • Humboldt (Germany)
  • La Sapienza, Rome (Italy)
  • Laboratoire de Robotique de Paris (France)
  • Melbourne/RMIT (Australia)
  • McGill (Canada)
  • Osaka (Japan)
  • Pennsylvania (USA)
  • Team Sweden (Sweden)
  • Tokyo (Japan)
  • UNSW (Australia)

10
The Scores
  • Pool 14 - 0 McGill (Canada)
  • QF 11 - 0 Humboldt (Germany)
  • SF 12 - 1 CMU (USA)
  • F 10 - 0 LRP (France)

11
Architecture
Behaviours
Visual objects
Role
Motor positions
Skill
Skill
Skill
Skill
Skill
Skill
Skill
Action Parameters
Robot x,y,q ball x,y
Vision, Object Recognition
Action Head Legs
Localisation
Move x,y,q
Visual Object dist, head
Camera
Motors
Robots, Ball and Soccer Field Environment
12
Robot Vision
13
Training robots to see colours
14
3D YUV Colour Classes
15
Colour Calibration
16
Object Recognition
17
Doggie Cam
18
What the robot sees
19
World Model
20
Localisation
(x, y)
d
New position after taking into account distance
to landmark
h
Current robot position according to world model
21
Localisation
(x, y)
d
New position after taking into account distance
to landmark
h
Updated position
22
Locomotion
23
Turning
24
Sideways Walk
25
Circling Behaviour
26
Conan Strategy
Goalie
Ball Lost
Find Ball
Low CF
Yes
Localise
Yes
No
Low CF
Localise
Forward
Yes
Yes
Find Ball Pos 100,5
Ball Lost
No
No
Ball close
No
Yes
No
follow at 50cm rad from (100,0)
Ball close
Upfield
Ball robot
Downfield
No
Yes
Aligned(R)
Yes
Rob too far
Back-Up
Go Side
Go Behind
Yes
No
Align
Dribble
Butt(R)
Dribble
27
Field Regions
6
3
5
4
1
2
28
The Arena
29
UNSW vs LRP
30
(No Transcript)
31
2001 a new robot
32
Whats next?
  • Radio Ethernet Team cooperation
  • Learning new behaviours
  • Improved colour calibration
  • Better robot recognition
  • Better dribbling, kicking, walking
  • Improve localisation
  • Anticipating ball trajectory
  • Switching camera resolution
  • Force feedback
About PowerShow.com