Intelligent Robotics and Embedded Systems

1
Intelligent Robotics and Embedded Systems Dr.
Marek Perkowski and Dr. Douglas
Hall http//web.cecs.pdx.edu/mperkows/
http//www.ece.pdx.edu/People/Hall1.htm
Dr. Hall
Robot Theatre as a Metaphore for robot-human
interaction and learning.
Building Interactive Robot Theatre using various
human-robot interaction and advanced robot
imitation methods is a long-term dream of
Professor Marek Perkowski, director of
Intelligent Robotics Laboratory. Their works
attempt to blend scientific theories, advanced
computational intelligence programming and
artistic insights to create robot actors for the
theatre. Unlike current robot theatres that are
fully programmed and have simplified robots or
standard mobile robots, the Portland Cyber
Theatre uses stationary and mobile robots that
are humanoid and that improvise during their
interactions with humans. Supported by grants
from Intel, KOSEF and Korea Advanced Institute of
Science and Technology, the group built the first
version - KAIST Hahoe Robot Theatre. The masks of
Sonbi a Confucian Scholar, Yangban a proud
aristocrat, and Pune his concubine, are
familiar to Koreans and to everybody interested
in history of theatre. Hahoe play is known in
Korea since 13 century. Now these characters
become animated actors, humanoid robots with
human-like behaviors, gestures and reactions.
Dr. Perkowski smiles at the Sonbi Robot Head
from Hahoe Theatre.
The ultimate goal of the Portland Cyber Theatre
is to develop the artistic concept of robot
puppetry based on sound human-robot interaction
principles and thus help to develop a general
theory of human-robot interaction for future home
robots that will become our life partners. For
instance, we investigate methods to teach robots
from examples and interaction, by imitation and
by playing human-robot games. How can a certain
expertise be transferred easily and flexibly from
a human or a team of humans to a humanoid robot?
We believe that future robots used in daily life,
and especially for the disabled and elderly will
need to address all or at least many of these
issues in order to be not only accepted, but also
be liked by their human owners.
To reach the above long-term goals, our research
concentrates on several separate issues such as
mechanical design of robots, new sensing and
actuating technologies, robot vision, natural
language based conversation and interaction,
gesture, body language and facial gesture
recognition, design of expressive theatrical and
emotional movements and automatic
learning/creating of them, cognitive modeling,
human-robot interaction and robot imitation, art
of animating movies and puppets, machine
learning, evolutionary hardware, quantum circuits
synthesis for robot behaviors, game theory and
quantum games. Our research is thus very
multi-disciplinary and we collaborate with groups
at PSU and worldwide on particular narrower
topics.
Inside the head
Our robots include robot arms, large humanoid
robots, walking hexapod, quadruped and biped
(humanoid) robots, robot heads (research of Dr.
Perkowski) and robotized wheelchairs (research of
Dr. Hall).
Sonbi the Confucian Scholar smiles back. He can
track you also with his eyes and head.
Selected Publications in Embedded Systems D. V.
Hall, Microprocessor System Design- Hardware,
Programming, and Interfacing, McGraw-Hill Book
Company, 2006. C. H. Lee, M. A. Perkowski, D. V.
Hall, D. S. Jun, "Self-Repairable EPLDs II
Advanced Self-Repairing Methodology," 2001
Congress on Evolutionary Computation, 2001. C. H.
Lee, D. V. Hall, M. A. Perkowski, D. S. Jun,
"Self-Repairable GALs," Journal of Systems
Architecture, 2001. C. H. Lee, M. A. Perkowski,
D. V. Hall, D. S. Jun, "Self-Repairable EPLDs
Design, Self-Repair, and Evaluation Methodology,"
Second NASA/DoD Workshop on Evolvable Hardware,
2000.
We are inspired by scientific studies of human
and animal behavior which demonstrate remarkably
clever mechanisms of predicting behaviors of
others and ways to learn through interaction. We
try to incorporate these mechanisms in our robots
not only for entertainment purposes or ways of
interacting with future humanoid robots but also
in an attempt to understand better the humans way
of learning and communicating. Sonbi can think
Is Marek smiling?, how is he smiling?, why
he is smiling when he tells me that he likes me?
These are difficult questions for a human, but
even more so for a robot that is using Hidden
Markov Model based face recognition combined with
Constructive Induction pattern recognition
method.
Selected Publications in Intelligent Robotics T-W
Wang, M. Sajkowski, T. Stenzel, M. Perkowski, "An
Inexpensive Educational Platform to Teach
Humanoid Robotics," Proc. ICEE conference,
Gliwice, Poland, June 2005. M. Perkowski, T.
Sasao, J-H Kim, M. Lukac, J. Allen, S. Gebauer,
"Hahoe KAIST Robot Theatre Learning Rules of
Interactive Robot Behavior as a Multi-Valued
Logic Synthesis Problem," Proceedings of the 35
th International Symposium on Multiple-Valued
Logic, May 19-21, 2005, Calgary, Canada, 2005.
S. Grygiel, M. Zwick, M. Perkowski, "Multi-level
decomposition of probabilistic relations,"
Kybernetes The International Journal of Systems
Cybernetics, Vol. 33, Number 5/6, pp. 948961.
ISSN 0368-492X, 2004. A. N. Al-Rabadi, M.
Perkowski, M. Zwick, "A comparison of modified
reconstructability analysis and Ashenhurst-Curtis
decomposition of Boolean functions," Kybernetes
The International Journal of Systems
Cybernetics, Vol. 33, Number 5/6, pp. 933-947,
ISSN 0368-492X, 2004. P. Burkey, M. Perkowski,
"Efficient Decomposition of Large Fuzzy Functions
and Relations,'' Proceedings of International
Conference on Fuzzy Information Processing,
Theories and Applications, March 1-4, 2003,
Beijing, China, Tsinghua University Press and
Springer, pp. 145-154, 2003. M. Folgheraiter, G.
Gini, M. Perkowski, M. Pivtoraiko, "Blackfingers
a Sophisticated Hand Prosthesis,'' Proceedings of
ICORR 2003 (the 8th International Conference on
Rehabilitation Robotics), April 22-25, 2003,
KAIST, Korea, pp. 238-241, 2003. M. Perkowski,
T. Sasao, A. Iseno, U. Wong, M. Pivtoraiko, M.
Folgheraiter, M. Lukac, D. Ng, M. Fix, K. Kuchs,
"Use of Machine Learning based on Constructive
Induction in Dialogs with Robotic Heads,''
Proceedings of ICORR 2003 (the 8th International
Conference on Rehabilitation Robotics), April
22-25, 2003, KAIST, Korea, pp. 326-329, 2003.
What is new? It is well known that standard
computer science models such as logic functions,
fuzzy logic, finite state machines, decision
diagrams, or neural nets are excessively used in
robotics to model movement, learning, cognition
and perception. The innovative component of our
research is to generalize all these classical
models to their corresponding quantum
computational intelligence models. Our
constructive induction method learns thus from
examples not only Boolean circuits but also
Quantum Circuits. The circuit learned from
examples is the specification of certain robot
behavior, it can be a motion or a way of
language-guided synthesized speech and response.
We hypothesize that the robots brain should be
quantum and thus we analyze how best to use the
concept of quantum mechanics to build its
components. Thus we use theories of constructive
induction, control and quantum learning which are
both new and not used so far in the area of
robotics.