39
Design of Intelligent Robotics René van de Molengraft December, 6 th , 2010

Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

  • Upload
    others

  • View
    6

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

Design of Intelligent Robotics

René van de Molengraft

December, 6th, 2010

Page 2: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

Theme

/ faculteit werktuigbouwkunde PAGE 211-12-2009

• Health: Care and Cure

Page 3: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

Mission

/ faculteit werktuigbouwkunde PAGE 311-12-2009

• Help elderly people to stay in their homes• Improve quality of surgery for surgeon and patient

Page 4: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

Research challenges

/ faculteit werktuigbouwkunde PAGE 411-12-2009

Learning• Low-level:

3D-sensing, world modeling, grasping, locomotion• High-level:

cognition, planning, coordination of multiple robots

Approach• Mixed Model/Data-based• Use priors

Page 5: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

Robocup@Home

/ faculteit werktuigbouwkunde PAGE 511-12-2009

Page 6: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

RoboCup use cases

/ faculteit werktuigbouwkunde PAGE 611-12-2009

Midsize league soccer• Scenario: structured, but highly dynamic• Low-level: world modeling, skills• High-level: strategy, coordination

@Home league• Scenario: static, but highly unstructured• Low-level: 3D-sensing, object recognition, grasping• High-level: cognition, learning

Page 7: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

RoboEarth Project

• Four-year funding from European Commission (FP7)

• Starting date: December 1st, 2009

• Consortium consisting of:

PAGE 712-03-2010

Page 8: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

Motion Control

• Performance• 90% feedforward

• 10% feedback

PAGE 812-03-2010

• Humans in day2day tasks• 90% feedforward – learning and memorization

• 10% feedback – sensory feedback

Page 9: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

Industrial robots

PAGE 912-03-2010

• Structured environment• Same scene, same task• No humans

Page 10: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

Domestic and care robots

PAGE 1012-03-2010

• How to deal with a complex world?• Changing scenes• Many possible scenarios

Page 11: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

RoboEarth

• How to deal with a complex world?• Changing scenes

• Many possible scenarios

PAGE 1112-03-2010

• RoboEarth – learning and memorization

• Share any reusable knowledge between robots

• World-Wide-Web for and by robots

Page 12: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

RoboEarth

PAGE 1212-03-2010

Page 13: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

Safety and risks

PAGE 1312-03-2010

• Consistency of data in RoboEarth

• Who is allowed to sharing what?

• Use existing protocols (HTTPS)

• Protect “personal” data (compare Streetview)

• Certified users

• Log and analyse usage

• Liability in case of malfunction?

• Autonomous operation

• Who is responsible?

Page 14: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

• “By the year 2050, develop a team of fully autonomous humanoid robots that can win against the human world soccer champion team”

• Joint development of technology in robotics and artificial intelligence

• Since 1997

Robocup

Page 15: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

• Founded spring 2005

• 20 active members

• 2nd place World Cup 2008/2009/2010

Tech United

Page 16: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

• fully autonomous

• vision and localisation

• fast and safe

• ball-handling (don’t clamp!)

• ball-kick

• team-play

Challenges

Page 17: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

• design a team of autonomous soccer robots

• robot is of DH/CS type

• team is of DH/DS type

• soccer = strategy + skills

Design goal

Page 18: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

• robustness

• performance

• time-to-next-tournament

• safety

• maintenance

• cost

Design criteria

Page 19: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

• FIFA rules

• additional Robocup rules

• size, weight, color

Constraints

Page 20: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

• manage complexity

• keep it simple

• hierarchy

• make choices explicit

• use models

• create robustness

• redundancy

Design rules of thumb

Page 21: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

team

robot robot robot

Bottom-up design

Page 22: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

• localisation

• soccer skills

• intercept, dribble, kick

• soccer roles

• attacker, defender, goalkeeper

• soccer tactics

• pass, assist, play system

Requirements

Page 23: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

• low level (hardware) functions

• drive

• kick

• ball handling

• sense environment

• (high) level (sofware) functions

• localisation

• skills

• roles and strategy

Identify functions

Page 24: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

HW

HW

HW

HW

HW/SW

HW/SW

HW/SW

HW/SW

SW SW

SW

SW

HW HW HW

HW/SW HW/SW HW/SW

SW SW

SW

SW

HW/SW

HW

System architecture

Page 25: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

• which sensors?

• vision

• odometry

• accelerometers

• laser

• hardware or software?

• cpu power

• sensor fusion

Design space exploration

Page 26: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

• vision

• number of cameras, type of camera, resolution, omnivision, mirror, construction, acquisition, feature extraction, modelfit, object tracking etc.

Localization

Page 27: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

• Interdisciplinary

• mechanics

• electronics

• software

• Approach

• model-based design (simulation and prototyping )

• early integration

Specialists needed

Page 28: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

Page 29: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

Page 30: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

• hardware

• omnivision camera+mirror

• electronic compass

• odometry with motor encoders

• software

• white-line detection

• orientation calibration

• sensor fusion

Our localization design

Page 31: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

White-line detection

Page 32: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

Evolution and revolution

/ faculteit werktuigbouwkunde PAGE 3211-12-2009

Devbot0 2005

Model 2006

Model 2007

Model 2008

Model 2009

Page 33: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

Page 34: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

Page 35: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

Page 36: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

Page 37: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

?

How to design a system of systems?

Page 38: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

• 5 agents: distributed system

• agents must be self-reliant

• minimise interaction/sync

• maximise cooperation

• aggressive environment

System of systems

Page 39: Design of Intelligent Robotics · RoboEarth •How to deal with a complex world? • Changing scenes • Many possible scenarios 12-03-2010 PAGE 11 •RoboEarth – learning and memorization

/ faculteit werktuigbouwkunde

• April 2011, German Open, Magdeburg

• July 2011, World Cup, Istanbul

• Dedicated goal keeper

• Robustified ball handling

• Passing during gameplay

Towards Istanbul, 2011