RobotCub

1
RobotCub

• IST-004370

www.robotcub.org
2
Main Keywords

• Cognition (manipulation)
• Human Development
• Embodiment
• Community Building

3
Two Goals or a two-fold Goal?
Advance our understanding of several key issues
in cognition
Create a physical platform for embodied cognitive
research
4
Main Figures

• RobotCub is a 5 years projects
• Presently the consortium is composed of 16
  Partners, 11 from Europe, 3 from Japan and 2 from
  the USA.
• Total Funding is 8.5 M and total effort is 1,651
  person-months (138 person-years)
• Coordinator is the LIRA-Lab of the University of
  Genova Giulio Sandini, Giorgio Metta and David
  Vernon

5
Current Consortium
6
An open system!
Robot-Cub engineering goal is to build a humanoid
platform whose design is open to the scientific
community and can be duplicated and improved by
the community of its users.
Clearly stated in our Consortium Agreement
We are looking for other projects willing to join
RobotCub with the same open attitude.
7
Seeking International Collaboration
The European Commission is funding part of the
internationalization activities and the project
will support the research of groups interested in
using the robot platform (about 2 Meuro have been
set aside for this purpose).
reference persons name from each of the ongoing
projects interested to contribute
8
Main Objectives (60 months)

1. ..a professionally documented, reproducible open
   platform shaped like a child-humanoid
2. ..the understanding through real-world
   implementation of exploratory and
   manipulation-based cognitive skills
3. ..the study and implementation of the initial
   period of human cognitive development in an
   embodied artificial system
4. ..building of an international scientific
   community

9
which rest on several props

• Unity of cognition body and mind-ware co-design
• Requirements hands, dexterity, etc.
• Exploiting holistic design
• Complex bodies requires global optimization
• Cost effectiveness
• An open system can lead to a more efficient
  design/exploitation process

10
Science-wise

• Manipulation as a link between action and
  perception
• Manipulation from grasping to gesture production
  with communicative intents
• From tool use to the creation of tool-ware

11
General layout (60 months)
Time Activities
0 Start of activities on specific aspects of cognition (several parallel activities), start the design of the CUB, start internationalization activities, start open source support
18 months First evaluation of cognitive investigationMechanical design of the CUB
3 years Robot-cub and mindware integrationLaunch of competitive call for proposals
3 years New investigation starts using the open platform
5 years Integration of the mindware developed during the five years
12
Activity allocation
Building the robotic platform
Doinga lot of science
Buildinga community
13
Activities
ACT 1 Mind (Cognitive Neuroscience)
ACT 2 Body (Cognitive Robotics)
ACT 3 Open System (Dissemination)
RC 3.1 Mechanical Drawings
RC 1.1 Eye-Head-Hand Coordination
RC 2.1 Head-eye and Vision
RC 3.2 Electronic Drawings
RC 2.2 Arm-Hand and Touch
RC 1.2 Bimanual Coordination
RC 3.3 Firmware and Software
RC 1.3 Interaction and Affordance
RC 2.3 Spine-Legs and Walk
Community Building
Learning and Development
RC 3.4 CUB Manufacture
RC 2.4 Electronics and Control
RC 1.4 Interaction and Imitation
RC 3.5 Internationalization
RC 1.5 Interaction and Communication
RC 2.5 Enabling Technologies
RC 3.6 Maintenance and Update
14
Child-like, how much?
243mm
369mm
Approx 934mm
439mm
Avg. 14Kg - 30.8 lb
15
Started playing w/ components
16
Temporal Dependencies (60 Months)
ACT 3 Open System (Dissemination)
ACT 2 Body(Cognitive Robotics)
ACT 1 Mind (Cognitive Neuroscience)
time
Define hardware architecture
Define manipulation experiments
Specifications
Experiments with existing platforms
Design new components
Design Documents
18 months
Scientific Dissemination Activities
Experiments with new components
Realize new components
Executive Documents
Realize whole system
Experiment with whole system
Revised Executive Documents
Robot-Cub Platform
Final Demonstration
Maintenance and Update
60 months
17
The six objectives of month 18
SO-1 A model of human infants cognitive
development based on recent and well documented
experimental results. The model shall include, in
an experimentally reproducible way, a description
of how the robotic artifact develops over time
showing the formation of manipulation skills of
varying levels. Psychophysical and behavioral
experiments will be planned and carried out to
answer specific questions on the implementation
and to acquire relevant missing information about
the developmental process.
Responsibility Claes von Hofsten (UNIUP)
18
SO-2 The complete design of all CUB components
and a suitable integration plan. This includes
the definition of the functional and technical
specifications of the CUB mechanics, electronics
and software architecture. At this stage of the
project, the partners responsible for the CUB
design and for testing the CUB individual
components will have a) completed the design
stage, b) run a mechanical compatibility test and
c) written a preliminary plan for integration. .
Responsibility Giorgio Metta (UGDIST)
19
SO-3 The initial results of the implementation of
cognitive abilities in an artificial system. This
objective will be demonstrated through extensive
testing of the robots cognitive abilities in
realistic situations, implemented in several of
the existing robotic platforms. In addition to
basic manipulatory and visual skills, the robots
will be equipped with a number of basic social
skills, enabling natural interactions between
robots and humans.
Responsibility David Vernon (UGDIST)
20
SO-4 Results of the testing of new technologies
to be used in the CUB platform. Particularly
important for the scope and goal of Robot-cub is
to monitor and to test continuously new
technologies for sensors and actuators as well as
the electronic (HW/SW) components of the Open
System Platform .
Responsibility Paolo Dario (SSSA)
21
SO-5 The Community Building activities will be
carried out all along the projects duration.
Responsibility Giulio Sandini (UGDIST)
22
SO-6 Define and refine the IPR and licencing
rules of robotcub
Responsibility David Vernon (UGDIST)
23
Contacts

• http//www.robotcub.org

Next meeting is November 25-27 in
Genova Everybody is welcome!
24
How do we get there through a roadmap?
25
Synthesize our knowledge of human cognitive
development

• Neuroscience leads ?
• Organize state of the art understanding of
  cognition
• Prepare, think, plan, new experiments to fill the
  gaps
• Derive suggestions for robotic experiments, CUB
  requirements, materials, skills, control, etc.

26
Since were focusing on manipulation
Development Development Learning what Perception/Action exploitation Component of Social Interaction Goal of system
Immediate No reaching yet Head-eye coordination Gazing, smooth pursuit Shared attention Look around
Immediate Pre-Reaching Approach an object Controlling arm and hand movements in space Pointing Touch
Delay between action onset and consequences Power grasping Eye-hand coordination based on object position and object motion Anticipatory closing of the hand Reaching for object held by other person I got it! Grasp (become owner)
Delay between action onset and consequences Differentiated grasping Adjustments to object shape and size Eye-arm-hand coordination based on object s shape Take and give Grasp appropriately (geometric)
Delay between action onset and consequences Object manipulation Objects affordances Eye-arm-hand Coordination based on actions to be executed on objects Play games Handle objects appropriately (use)
Long delays Imitate acts on objects Associate what is seen with what the system can do What I do looks like what I see Play games Actions Interpretation
Long delays Act to Communicate Associate what is seen (perceived) with meaning What I do generates some reactions Communication Actions meaning
27
Start the robot design investigation

• Leveraging on our past experience different
  aspects of the CUB design are handled in parallel
• Specific targeted investigation on, for instance,
  materials or new sensing, actuation technology
  might be launched budget has been allocated for
  this purpose

28
In parallel start mindware investigation

• Robotic experiments are initiated relying on
  existing setups
• Aims
• clarify specific aspects of interests along the
  developmental timeline
• Uncover the developmental mechanisms useful in
  building adaptive machines
• Clarify the link between action, perception, and
  cognition

29
Opening the system

• Starting the organization of the open system
  repository
• Establishing adequate level of documentation
  both SW and HW
• Establishing standards (e.g. for sharing
  upgrades)
• Attracting the community to the CUB

30
Start building a community

• Dissemination advertise the project worldwide
• Training getting young scientists involved with
  cognitive robotics
• Lobbying getting other funding agencies (third
  country participants) involved with financing
  cognitive robotics actions

31
_at_ month 18

• Coalesce the activities into
• Design of the CUB and integration plan
• Design of parts of the mindware architecture and
  stricter integration plan
• Specific results from technology search to go
  into the CUB, from neuroscience to fill the
  gaps, and from the internationalization and
  training activities international community
  building
• and more importantly a demonstration!

32
The vision _at_ 18 months

• Robot ? Play-pal
• Sitting around a table, playing with toys, tools,
  objects
• Learn how to
• Properly grasp a toy/tool
• Associate tools to specific actions
• Adjust the action wrt the environmental
  conditions
• Combine primitive grasp/manipulative actions
• Interact appropriately with the play-pal in
  learning a new game or task

33
Milestones _at_ month 18

• Initial specifications of the CUB (open system)
• Definition of the cognitive architecture and
  initial implementation of manipulative behaviors
• Creation of the core components of the
  international community and plans for an
  international project