Tennessee State University College of Engineering

grandgoatΤεχνίτη Νοημοσύνη και Ρομποτική

23 Οκτ 2013 (πριν από 3 χρόνια και 11 μήνες)

80 εμφανίσεις


Tennessee State University

College of Engineering


ENGINEERING RESEARCH INSTITUTE (ERI)

Interdisciplinary Research in Robotics


Intelligent Tactical Mobility Research Laboratory (ITMRL)

Intelligent Control Systems Laboratory (ICS)

Center for Neural Engineering (CNE)

Computer and information systems Laboratory (CISE)

Mohan J.
Malkani
, Ph.D. (Director)

(615) 963
-
5400 Fax: (615) 963
-
5397

mmalkani@tnstate.edu



Research Projects in Robotics: Past and Present



Tele
-
Robotics

jointly

with

Caltech

funded

by

NSF

(
1997
-
2000
)



Originally

Funded

by

US

Army

TACOM,

Warren,

MI,

under

two

research

grant

contracts
:



1
.

Development

of

an

Integrated

High
-
level

Mobility

Controller

for

Virtual



Tandem

Robotic

Vehicles,

DAAE
07
-
98
-
C
-
0029
,

(
1997
-
2000
)



2
.

Deliberative,

Reactive,

and

Adaptive

Task

Planning

of

Intelligent



Cooperative

Mobile

Robots,

DAAE
07
-
01
-
C
-
L
-
065
,

(
2001
-
2002
)





Embodiment

of

intelligent

behaviors

on

mobile

robots

using

fuzzy
-
genetic

algorithms,

funded

by

NASA/Ames

Research

Center

(
2000
-
2004
)


Funded

by

DARPA

Through

Penn

State

Applied

Research

“Sensor

surveillance”

under

MURI
-
ESP

Research

Project,

DAAD
19
-
01
-
1
-
0504
,

(
2002
-
2003
)



Funded

by

NASA/JPL,

FAR

Investigator

Program,

“Visual

Telerobotic

Task

Planning

of

Cooperative

Mobile

Robots”,



(
2003
-
2006
)


Development of Advanced Control schemes that enable tactical team
cooperation of Intelligent Autonomous robots effectively and efficiently.


Test and evaluate performance of advanced control schemes under
different operational conditions and different sensory data modality
experimentally using high
-
fidelity computer generated simulation and
physical robotic test beds.


Technical Competency Areas Included:


Behavior
-
based Distributed control of Cooperative Mobile Robots.


Sensory data and image processing and fusion for fault tolerance
control of intelligent robots.


Advanced control schemes based Soft Computing techniques, (Neural
Networks, Fuzzy Logic, Genetic Algorithms, …).


High
-
fidelity world perception modeling of robotic systems.


Man
-
machine development for Visual Teleoperation and Telerobotic
control of Cooperative Robots.

Research Focus Areas




Developed various behavior
-
based schemes for
intelligent deliberative, reactive, and adaptive
task planning of cooperative robots.


Developed various image processing techniques
for visual localization and target tracking of
robots.


Applied different soft computing methods for
target pattern recognition and classification.


Developed FMCell comprehensive robotic
simulation software for the purpose of man
-
machine interface development.


State
-
of
-
the
-
art physical robotic test bed
consisting of twelve heterogeneous robots.


Embodiment of intelligent behaviors on mobile
robots using fuzzy
-
genetic algorithms


Theoretical and Experimental

Research Capabilities

Intelligent Man
-
Machine Interface



Interactive Component Based Architecture for
rapid task deployment of cooperative robots.


Image and sensory data processing and analysis
capability for intelligent control of autonomous
robots.


Soft computing capability for deliberative,
reactive, and adaptive development of behavior
-
based robot tactical schemes.


3D modeling and simulation tools for world
perception modeling and visualization of
cooperative mobile robots.


Built
-
in TCP/IP wireless communication
protocols for distributed client/server
-
based
control of remotely operating robots.


Experimental human
-
robot interaction



Robotics Research

Human
-
Robot Interaction

Intelligent Control Systems

Telerobotics

Intelligent User Interfaces

Multi
-
Robot Cooperation

Interoperability

Software Architectures




Human
-
Robot Interaction


Over the Internet; Via PDAs; Via Speech


Via cellular phones (speech integrated)


Human detection, recognition, and localization


Social behavior modeling Interoperability for
Robotics


Programming language and operating system
independent software architecture



Intelligent User Interface Design


Adaptive
-

mission aware


Multiple users


multiple robots



Heterogeneous Multi
-
Robot Cooperation


Behavior
-
based approach

Robotics Research

The Human Agent System


The Human Agent
i
s a virtual agent
that serves as an
internal
active
representation

of
people in the
robot’s
environment.


As a
representation,

it is able to detect, represent and monitor people. The description
active

is used, much
as in describing active perception vision systems [Bajcsy 1987], to indicate that the system can take action
to make its representation richer.



Human

Detection

Agent (motion)

Human

Detection

Agent (sound)

Affect

Estimation

Agent

Human

Identification

Agent (face)

Human

Identification

Agent (voice)

Human Database

Identification

Agent

Human

Affect

Agent

Observer

Agent

Monitoring Agent

Human

Intention

Agent

Social

Agent

Interaction Agent

Human Agent

To Self

Agent

Sensory EgoSphere (SES) for Mobile Robots


Peters redefined the
Sensory EgoSphere as a
sparse spatiotemporally
indexed short term memory
(STM).



Structure: a variable
density geodesic dome.



Nodes: links to data
structures and files.



Indexed by azimuth,
elevation and time.



Searchable by location and
content.

images

sonar

laser

Peters, R. A. II, K. E. Hambuchen, K. Kawamura, and D. M. Wilkes, “The Sensory Ego
-
Sphere as a Short
-
Term Memory for Humanoids”, Proc. IEEE
-
RAS Int’l. Conf. on Humanoid Robots, pp. 451
-
459, Waseda
University, Tokyo, Japan, 22
-
24 Nov. 2001.

Experimental Design


2 training tasks with the
original and enhanced
interface


2 teleoperation tasks with
the enhanced and orignal
interface.

Telepresence Software Architecture (Over Internet)



Robot Control Programs



Internet Control (ServerSide)



Internet Control (Client

Side)

TCP/IP
(Internet)

USER


SERVERS

API



Hardware

Human

Commander

Robot

Commander

Audio
Commands

Soldiers

Speech
Recognition

TCP/IP

Internet

Research Motivations
(
Consumer Tele
-
presence)


MANAGER

Robot
-
1 Grabs


an Image

Process

(NN
-
Fuzzy)

Grab

Robot
-
2 Grabs


an Image

Process

(NN
-
Fuzzy)

Image

Image

Grab

Robot
-
3 Grabs


an Image

Process

(NN
-
Fuzzy)

Image

Grab

Final Decision

(Fuzzy Logic)

Fuzzy Decision
-
2

Fuzzy Decision
-
3

Fuzzy Decision
-
1

System Architecture

Research Motivations
(
Development of Robot Behaviors, NASA, Phase
-
I)

FIRBA Implementation


Abstracts beeSoft:

Complex API protocols
are hidden


Object Oriented:

Abstraction, reuse by
inheritance.


Perception Sharing:

Common perceptions can
be shared


Action Suggestions:

Arbitration through MAL,
fuzzy inference and De
-
fuzzification.


Independent Behaviors

Overall Software Architecture.

LEVEL 1 BEHAVIORS

LEVEL 0 BEHAVIORS

SONAR

HANDLING

SONAR

CLASS

ODOMETER

HANDLING

ODOMETER

CLASS

SONAR PERCEPTIONS

TARGET

HANDLING

TARGET

CLASS

PATH

HANDLING

PATH

CLASS

TARGET PERCEPTIONS


MOTION

PRIMITI
-
VES

Research Motivations
(
Development of Robot Behaviors, NASA, Phase
-
I)


SENSORS

Pre
-
Perception Processing

Perception Capabilities

Behaviors.

Action Capabilities

Action Execution

ACTUATORS

The FIRBA architecture.

FIRBA


Robot Control System


Complexity
-



Robustness



Multiple Sensors



Multiple Methods



Integration



Incremental Development



Software


This complexity is handled by system
decomposition in terms of :


--

functional units


--

behavioral units