Challenge 2 - RTD

spectacularscarecrowΤεχνίτη Νοημοσύνη και Ρομποτική

17 Νοε 2013 (πριν από 3 χρόνια και 10 μήνες)

76 εμφανίσεις

•••
1

Challenge 2

Call 3


presentation to NCPs

Brussels, December 13, 2007



Colette Maloney, PhD

Head of Unit, INFSO E5, Cognitive Systems and Robotics

European Commission

•••
2

Challenge 2: Cognitive Systems, Robotics, Interaction


Backdrop



the proliferation of sensors in IT applications (robots,
computer vision, speech recognition) and their deployment in

real
-
world’ settings
;




the consequent requirement that information processing be
capable of dealing with
uncertainty

(in the form of
unexpected events, diverse contexts, novel situations, etc)
inherent in sensory data;


•••
3

Target a (IPs, STREPs)

3 areas: focus on one

•••
4

Robots …


… in real world settings


less
-
constrained environments can be too nuanced, too complicated
and too unpredictable to be summarised within a limited set of
specifications










there will inevitably be novel situations and the system will have gaps,
conflicts or ambiguities in its own knowledge and capabilities

•••
5


robots handling different objects and operating
autonomously or in cooperation with people


may call for


manipulation & grasping,


navigation,


locomotion,


obstacle avoidance,


interaction with humans,…


•••
6

Monitoring and controlling…

eg using computer vision





traffic monitoring




other applications



intelligent surveillance,

biometric recognition,

exploration,

data
-
gathering

manufacturing,

robotics,…


•••
7

robots or other systems monitoring & controlling material or
informational processes

May call for:


detection,


recognition,


classification


…. of objects, events or processes,…


•••
8

Multimodal Interfaces …

eg in speech recognition improvements have come
from increases in computing power


the majority of mobile phones have voice dialling




software for dictating documents on your PC is available in
most computer stores




Interactive Voice Response (IVR) systems are becoming
commonplace for handling telephone enquiries


… technology is fragile in ‘real’ conditions


From Roger K. Moore
Spoken Language Processing for Artificial Cognitive Systems, IST 2006

•••
9

multimodal interfaces &/or interpersonal communication
systems understanding language, gestures, etc

may call for


a deep understanding of human physical &
cognitive capabilities, communication needs &
contextual constraints,…


•••
10

A new systems engineering paradigm?


as systems get more complex,
the
growth in utility and productivity is in
ever smaller proportion to the growth in compute power

afforded by
Moore’s Law




cognition
: a
way of overcoming uncertainty accompanying interaction in a
world that is not completely pre
-
specified



goal:
engineering principles

for robust, versatile systems



•••
11

A key question




How should systems pertaining to these areas be designed
and built

so that they are more robust, flexible, effective,
natural and where necessary or desirable, safer and more
autonomous than what is possible today?



but,… how can we specify
what it means to be robust,
flexible
, etc?

•••
12

Addressing the key question



while providing valid and viable answers to the previous key
question,
projects can take the approach of their choice
, and
draw on those scientific and engineering disciplines that are
needed to achieve the its goals.



projects

are

requested

to

contribute

to

the

development

of

criteria

for

benchmarking

system

properties

such

as

robustness,

scalability

and

adaptability,

make

them

public

and

compare

with

others


don’t forget the emphasis, in particular in robotics, is on
integration of complete systems

•••
13

Project outcome


projects

are

expected

to

develop

know
-
how

needed

to

create

new

products

and

to

build

systems

that

are

desirable

but

cannot

be

built

given

our

current

know
-
how




emphasis

is

scientific

and

technological

advance



not

application

development
;





role of applications is to provide
research questions

and
to demonstrate the impact of conceptual or technical
innovation.


•••
14

Challenges for research




a key issue is
how

systems should work


greatly improving
robustness etc. requires
rethinking

the way systems are
engineered



theories

are needed
-

systems theories, software
architectures, control theories, modelling theories, etc
-

that
will allow us to build these types of systems



engineering progress will depend on
advancing scientific
understanding

of what both
natural and artificial

systems can
and cannot do, and how and why



integration of disciplines
: robotics, artificial intelligence,
computer vision, natural language, cognitive science,
psychology,… mathematics,… philosophy

•••
15

Target b (NoEs)

•••
16


Network of excellence (NoE)

-

programmes of
joint research & resource sharing which contribute to
reinforcing & sustaining scientific excellence.


Robotics: experimentation with industry strength
platforms, benchmarking


Cognitive Systems: integration of diverse research
area, understanding of requirements for specific
applications


Language based communication and interaction: new
approaches; understanding of capabilities required
of technical systems

•••
17

Target c (CSAs)

•••
18

Coordination Action


Increased cooperation and coordination between EU
Member States covering domain that contribute to
overall goals of Challenge 2.

•••
19

Some myths


European level



does not mean having a spread of partners
from countries all over Europe.

It means cross
-
border
collaboration that promises to achieve more than could have
been achieved within one single Member State


Industrial participation

-

is not a requirement
. It is an option


Presence of one or more
SMEs

-

is not to be taken as a must or
as a de facto plus point
. SMEs are treated just as any other
partner in a consortium in terms of having a necessary
competence, a reputation and a clearly defined role


Projects
-

incl. IPs
-

need not assemble large numbers of
partners.
Projects should only include those partners needed to
achieve the goals and no more.

•••
20

More myths



Management by a professional consultancy


is not to be taken
for granted.
It must offer a proven added
-
value.


We are not looking for ideas for new applications systems or
products.

Advances should be related to the engineering goals
of Challenge 2


not to health, security, ambient
intelligence,….(those such domains can provide useful
demonstration scenarios)


We are not looking for ‘contributions’ to any policy except
research policy.

And the part of research policy in question is
Challenge 2
-

not enlargement policy, SME policy, Information
Society policy (i2010), other Community policies*….

*often cited wrongly or misunderstood

•••
21

Links


More (documents, project descriptions, presentations,…) at



http://www.cognitivesystems.eu


See also



http://www.eucognition.org



http://www.roboticsplatform.com


Contact



Colette.Maloney@ec.europa.eu



Cognitive Systems, Interaction, Robotics: 2007
-
08

•••
22








-

END
-