Emergence of Cognitive Grasping through Emulation, Introspection and Surprise

loutclankedAI and Robotics

Nov 13, 2013 (3 years and 6 months ago)

64 views



The aim
of GRASP
is the design of a cognitive system capable of performing grasping and manipulation
tasks in open
-
ended environments, dealing with novelty, uncertainty and unforeseen situations.


To meet the aim of the project, studying the problem of object manipulation and grasping will provide
a theoretical and measurable basis for system design that is valid in both human and artificial systems.


To demonstrate the feasibility of our approach, we will instantiate, implement and evaluate our
theories and hypotheses on robot systems with different embodiments and complexity.


GRASP goes beyond the classical perceive
-
act or act
-
perceive approach and implements a predict
-
act
-
perceive paradigm that originates from findings of human brain research and results of mental training
in humans where the self
-
knowledge is retrieved through different emulation principles.


The knowledge of grasping in humans can be used to provide the initial model of the grasping process
that then has to be grounded through introspection to the specific embodiment.


To achieve open
-
ended cognitive behaviour, we use surprise to steer the generation of grasping
knowledge and modeling.

1.
Theory of Grasp Modeling

-

We study the requirements and effects of the agent's embodiment
on the situatedness, awareness, task and environment understanding and thus provide the means
for adaptation and self
-
reasoning.

2.
Self
-

and Context
-
Awareness

-

We investigate how an agent benefits from using tutor based
and autonomous exploration together with physical modeling of the world to learn more about
the possibilities and constraints offered by its embodiment.

3.
Curiosity and Surprise Driven Behaviour

-

We
will
show how the detection of an unexpected
event or action is exploited to efficiently add new values, categories or dimensions to the grasping
ontology while at the same time exploiting surprise to bootstrap the learning process.
Expectations are derived from the prediction of agent behaviour using the experiences gained
from self
-
awareness or introspection.

4.
Inferring new Grasping Strategies

-

We will use the ontology and acquired general
knowledge to generate expectations for grasping and manipulation tasks as means of correction
between the predicted and the actual state. This will allow adaptation to new objects and
situations without the need for extensive re
-
programming.

5.
Exploitation to Future Prosthesis, Industrial and Service Markets

-

W
e set out to exploit
the theoretical findings by investigating the grasp mapping to different artificial hands. The
objective is to learn how kinematical design and the number of DOFs influence dexterity and how
to optimize the graspable sub
-
set of all possible grasps while minimizing DOFs.

1.
Learning to Observe Human Grasping
and Consequences of Grasping

2.
Representations and Ontology for
Learning and Abstraction of Grasping


3.
Self
-
experience of Grasping and
Multimodal Grounding

4.
Perceiving Grasping Context and
Interlinking Contextual Knowledge

5.
Surprise: Detecting the Unexpected
and Learning from it

6.
Introspection and Prediction through
Simulation

7.
Cognitive Robotic Grasping
-

Integration and Applications

RESEARCH APPROACH AND WORKPACKAGES

PROJECT OBJECTIVES

PROJECT INFORMATION

CONSORTIUM

COORDINATOR:

Kungliga Tekniska Högskolan

Stockholm, Sweden

Dr. Danica Kragic

Dr. Patric Jensfelt

http://www.nada.kth.se/cvap

http://www.cas.kth.se

Universität Karlsruhe (TH)

Karlsruhe, Germany

Dr. T. Asfour

Prof. R. Dillmann

http://wwwiaim.ira.uka.de

http://www.uni
-
karlsruhe.de


Techniche
Universität

München

Munich, Germany

Prof. D. Burschka

http://www6.in.tum.de/burschka/

Ludwig
-
Maximilians
-
Universität

Munich, Germany

Prof. H. Deubel

http://www.psy.uni
-
muenchen.de/


Lappeenranta University of
Technology


Lappeenranta, Finland

Dr. V. Kyrki

Prof. H. Handroos

http://www.it.lut.fi/


Technische Universität Wien

Vienna, Austria

Prof. M. Vincze

http://www.acin.tuwien.ac.at/


Foundation for Research

and Technology
-

Hellas, Greece

Prof. A. Argyros

Dr. M. Lourakis

http://www.ics.forth.gr/


Universitat Jaume I

Castellón, Spain

Dr. A. Morales

Prof. P. Sanz

http://www.icc.uji.es/


Otto Bock

Vienna, Austria


Dr. H. Dietl

http://www.ottobock.at/


CONTACT:

Danica Kragic

www.csc.kth.se/~danik

Tel: +468 790 6729

Fax: +468 7230302

www.grasp
-
project.eu

Emergence of Cognitive Grasping through
Emulation, Introspection and Surprise