Center for Robot Intelligence

fangscaryΤεχνίτη Νοημοσύνη και Ρομποτική

13 Νοε 2013 (πριν από 3 χρόνια και 9 μήνες)

84 εμφανίσεις





Center for Robot Intelligence


















CONFIDENTIAL DRAFT










Center for Robot Intelligence

Carnegie Mellon University


CONFIDENTIAL

9

1

Vision

Bold missions of exploration, discovery and the development of space demand cognitive robots
and reliable autonomous spacecraft. Scenarios in which robots operate with

minimal Earth
communication are essential to this future and compel autonomy. We envision mission
successes through robots with competence that leaps beyond current state
-
of
-
art, and
perception and cognition capabilities that are manifested and evolved ov
er long, productive
robot lifetimes.

Robots will be our agents of planetary surface and deep space exploration, handling the
repetitive and time
-
consuming tasks of data collection and data reduction. Teams of robots will
survey vast regions, cataloging and

classifying geological features and formations and
searching for evidence of life, all the while tailoring their investigations to what they discover and
seeking opportunities to discover more. Robots will be humankind’s agents and partners in
space, cons
tructing and servicing orbital and surface facilities with millions of components. They
will operate on distant planets, using high
-
level directives, responding to and interacting with
humans.

Our approach to achieving this vision is research, education,
and manifestation of intelligent
space robotics through a Center for Robot Intelligence. The Center will perform breakthrough
research in robotics technology relevant to NASA missions and will deliver the next generation
of space roboticists.

2

Challenges

At

their current pace of evolution, robots will not achieve the requisite intelligence and
functionality to fulfill space missions of the next decades. Though computing power continues to
increase rapidly, we must address fundamental issues and make breakthr
oughs to realize
intelligent behavior.

Robots presently fall vastly short of needs and ambitions, unable to do all that NASA would like
them to do. Currently humans must help robots get work done rather than the other way around.
State of the art robot mis
sion operations (e.g., Pathfinder and NEAR) utilize expensive
infrastructure

engineers, scientists and deep space antennas

to command and monitor every
move. In complex servicing scenarios, robots like Station and Shuttle manipulators do not
approach the E
VA capability of astronauts. In inspection scenarios, robots like AERCam have
precise sensing ability but fall far short of human interpretation and judgment.

Robotics is currently perceived to be a high
-
risk technology. What robots do now, they do
without

requisite reliability. Robots have poor reasoning abilities, so tasks have to be scripted in
great detail

almost to the finest level of granularity. Robot software itself is brittle and
consequently robots cannot cope with unexpected circumstances or reco
ver gracefully from
failures. An equally important driver is the expense of robot development and system
integration. Even in relatively low
-
cost missions, software engineering is a significant portion of
the overall mission budget. It also represents too

large a portion of the system development
cycle, taking precious time away from system validation.

Overcoming these problems demands concentrated, focused efforts to create breakthrough
intelligent robotics technologies and systems, and to define and appl
y best available software
Center for Robot Intelligence

Carnegie Mellon University


CONFIDENTIAL

10

practices to intelligent robot development. The next generation of intelligent robots must be
robust to uncertainty and to changes in their environments. No longer will it be necessary for
designers to account for and model all de
tails of the robot’s world. These intelligent robots must
be flexible with respect to modifications of their task directives. They will understand their own
state, their level of performance, and the state of the world, and will reason about this
informati
on to plan courses of action. Intelligent robots will grow smarter with experience,
learning from their mistakes and learning to avoid mistakes in the first place.

3

Research in Robot Intelligence

Our proposed Center for Robot Intelligence will enable space
enterprise by creating
breakthrough robotic technologies and systems, and by training current and future NASA
researchers in the style of creative multi
-
disciplinary work characteristic of robotics. Intelligent
robotics as a field is poised for major advan
ces in the coming years. By assembling a critical
mass of researchers, with diverse, complementary skills and backgrounds, committed and
enabled toward the vision of intelligent space robotics, we will deliver the breakthroughs and
create the novel systems

that fulfill NASA’s future mission and manpower objectives in this
essential area.

Interfaces are key issues in robotics
: interfaces between different system components, and
among the different sub
-
disciplines of intelligent robotics. Through rapid protot
yping of robotic
systems, researchers will gain unique experience in designing and integrating complex robotic
systems that work in the real world. They will learn to design boldly, to build quickly, and to draw
on the varied skill sets of all members of a

community. Since the manifestation of intelligent
systems is much more than putting together high quality components, the Center will create
techniques for modeling the performance of components and systems, detecting failures and
compensating for them.
The Center will improve software engineering techniques and develop
new ways for robots to reason about goals so that robots can adapt to achieve those goals in a
changing, unpredictable environment.

The Center for Robot Intelligence will be sited at Carne
gie Mellon University, drawing on broad
excellence in AI, robotics, and software, and employing a collaborative research approach in the
educational environment. Innovative research necessarily occurs at the cutting edge. The
proposing researchers are fa
miliar with emerging technologies and attuned to the frontiers of
the field. Graduate students offer cutting
-
edge skill and ability, undistracted focus, and fresh
and enthusiastic inquiry in the intellectual exploration fostered by the university setting.

Together, the perspective is broad and deep on fundamental, theoretical and experimental
issues and less constrained by programmatic concerns. Dramatic innovations are possible with
the riskier and more revolutionary developments. These advantages will f
oster and drive the
Center’s accomplishment of groundbreaking research and developing the next generation of
skilled space roboticists.

Robotics is inherently multi
-
disciplinary, but too often, robotics research has focused narrowly
on component and applic
ation
-
specific technologies. Our approach is to concurrently research
both general concepts and specific problems in a complementary fashion. General concepts
expose those ideas that will transfer from one problem domain to another; specific problems test

general concepts for deficiencies and provide clues to additional general concepts. As part of a
unified approach, projects in different application areas will gain synergy by sharing common
solutions.

Center for Robot Intelligence

Carnegie Mellon University


CONFIDENTIAL

11

All of the Center’s activities are designed for rele
vance to exploration, discovery and
development of space. Close ties with NASA centers will ensure that we respond to genuine
needs of space mission developers. We have tentatively identified four broad applications in
which intelligent space robotics is
poised to have a major impact: surface exploration; in
-
space
exploration; surface operations; and in
-
space operations (Figure 1). These applications require
general capabilities; intelligent exploration robots must be capable of self
-
directed inquiry and

discovery while intelligent work robots must be capable of performing assembly, inspection and
maintenance tasks autonomously.

These capabilities have much in common. In particular, they share two overarching
requirements of
flexibility
and
robustness
.
Flexibility is to the ability of a system to achieve new
goals. Flexible systems are not confined to small niche operations; rather, they have a wide
Figure 1. A balance of applications and technologies shapes the Center

Flexibility

Reliability

Intelligent

Space Robotics

Inquiry

and

Discovery

Sur
face

Exploration

In
-
Space

Exploration

Assembly,
Inspection, and
Maintenance

Surface Work
Operations

In
-
Space Work
Operations

To tolerate and
accommodate
component failures

To use components in
different combinations for
different tasks

Robustness

Adaptability

Taskability

To learn about and respond
to new environments and
new tasks

Applications

Capabilities

Requirements

Technologies

Center for Robot Intelligence

Carnegie Mellon University


CONFIDENTIAL

12

variety of capabilities, some of which may not even have been foreseen by their designers.
Flexible syste
ms will be ideal for exploring unknown environments and interacting with humans
in complex and varied tasks. Robustness is a system’s ability to handle the unexpected. A
robust system can, for instance, continue performing its mission even if a component
fails or if
the environment changes in an unexpected way. Robustness is essential for long
-
duration
missions where human control and monitoring must be minimized.

The requirements of flexibility and robustness are satisfied with three fundamental technolog
ies:
taskability
,
adaptability
, and
reliability.
Taskability is the capacity for components of a system to
be re
-
tuned and combined in different ways for different tasks. Adaptability is the ability to learn
about new environments and new tasks. Reliabil
ity is being able to tolerate or accommodate
component and task failures.

There are important synergies between these technologies. For instance, automated planning
can be used to implement taskability, since it frees designers from specifying plans for e
very
possible situation, thereby enabling a wider range of tasks. However, the same planner can be
used to increase system reliability when it is applied to an internal model of the robot and used
to plan a sequence of recovery actions to, for example, de
al with a failed motor. We expect that
many component technologies will exhibit similar synergies.

The technologies of taskability, adaptability, and reliability cut across application boundaries
(Figure 2). Adaptability is as important to a Mars surfac
e exploration rover entering new terrain
as it is to an orbital facility assembly robot working in and out of shadows. Taskability is
necessary in any robotic system whose mission goals are likely to change during its lifetime; it
can enable more great su
ccesses like the extended exploration missions of Voyager 2 and
Galileo. Reliability is clearly central to all space missions, but even more so for the highly
autonomous robotic operations we and NASA envision.

In summary, the University Center for Robot I
ntelligence offers a unique opportunity to research
tightly inter
-
related technologies. We expect that in a close
-
knit, multi
-
disciplinary group, novel
and valuable ideas will germinate and spread rapidly, allowing the Center to make breakthrough
advances

far beyond those from a similar investment in more scattered research.

Figure 2. Technologies of the Center cut across domain boundaries


Multi
-
robot systems, reconfigurable robots, goal
-
oriented behavior,

human/robot i
nteraction, planning and scheduling


Automated validation, monitoring and diagnosis,

fault tolerance, software engineering


Machine learning, neural nets, genetic alg
orithms, Markov models,

adaptive control


Taskability

Reliability

Adaptability

Surface

Exploration

In
-
Space

Exploration

Surface Work
Operations

In
-
Space
Work
Operations

Center for Robot Intelligence

Carnegie Mellon University


CONFIDENTIAL

13

4

Education in Robot Intelligence

A primary goal of the Center is to provide NASA with personnel who have the skills to research
and apply intelligent robotics. We will develop an educat
ional program aimed both at students
who will eventually join NASA, and at current NASA professionals who need to immediately
develop expertise in intelligent robotics. In particular, the objectives of the Center’s education
program are:



to create a group
of degreed roboticists who specialize in the technologies of
intelligent space robotics;



to create a pipeline of Robotics PhD and MS degree recipients into NASA;



to foster widespread use of intelligent robotics technologies in NASA through off
-
campus Maste
rs degree and certification programs directed toward current NASA
employees; and



to maintain relevance and vitality of our education and research activities through
personnel exchange, professional workshops and seminars.

The Center will offer research
-
di
rected graduate degree programs at Carnegie Mellon and
course
-
only Masters degrees and certificates to off
-
campus students. The education program
will leverage on the existing educational and administrative infrastructure present at Carnegie
Mellon. The
basis of the curriculum will be a broad foundation in robotics, including control
theory, mechatronics, artificial intelligence, and software engineering. In addition, the program
will enable students to pursue a specialized focus in cutting
-
edge areas su
ch as computer
vision, machine learning, automated planning and scheduling, highly reliable computing, MEMS,
and bio
-
mimetic system theory.

Research will be the capstone of the on
-
campus graduate program. Students will pursue
independent research as col
laborators and peers on Center
-
directed projects. In addition to
developing critical abilities for independent thinking and experimental methods, the program will
seek to closely acquaint students with NASA’s technology challenges, its field centers and,
most
importantly, its people. The Center will promote student interaction with NASA researchers and
developers through internships at NASA centers and through exchange programs where NASA
researchers can work at Carnegie Mellon (and vice versa) for extend
ed periods of time.

The goal of the off
-
campus program will be to enhance the skill sets of active NASA employees,
allowing them to apply robotics technology to meet NASA mission objectives. The off
-
campus
program will target part
-
time participation by N
ASA employees who would be otherwise unable
to participate in the Center’s educational programs. It will offer the choice of either a Masters
degree or certification in intelligent space robotics. Both the degree and certification programs
will emphasize
course and project activities rather than research, and will be administered
remotely from Carnegie Mellon.

In addition to formal education, the Center will organize and host regular workshops and
seminars on intelligent space robotics. Attendees from NA
SA, the commercial space sector,
and academia will gather to present state
-
of
-
the
-
art advances in space robotics technologies
and chart the future of the field. These workshops will be a principal forum for the dissemination
of research findings stemming
from Center projects and from the field at large.

Center for Robot Intelligence

Carnegie Mellon University


CONFIDENTIAL

14

5

Qualifications

5.1

Technical Excellence in Space
-
Relevant Robot Intelligence
Research

The following paragraphs illustrate some of Carnegie Mellon’s current robotics research thrusts:
extreme terrain mobility, e
xploration autonomy,
and

robot worksystems
. The research themes
and technology developments of these projects provide springboards to the proposed Center’s
programs. Furthermore, the working relationships fostered through collaboration with NASA
personnel
provide the basis for stronger, broader interaction between the proposed Center and
the NASA community.

Extreme Terrain Mobility

Scientifically interesting destinations present challenges beyond the
difficulties surrounding contemporary lunar and Martian
landing sites.
Extreme terrains like canyons and volcanoes attract scientific inquiry;
they source heat and water that may foster undiscovered forms of life,
and provide clues to the interior processes of moons and planets.

Dante’s

technologies enable gath
ering of scientific data from such
extreme environments, relevant on Earth and beyond. Spherical laser
scanners, trinocular stereo vision, capaciflective foot sensors, and
task control software allow robot explorers like
Dante

to plan
traverses. Complemen
ting that machine intelligence, novel terrain
-
accommodating locomotion and spider
-
like tethering afford
unprecedented mobility in steep, soft, boulder
-
strewn terrain.

In the summer of 1994,
Dante

descended into the Mt. Spurr volcano
in Alaska, gathering da
ta on carbon dioxide, hydrogen sulfide and
sulfur dioxide content in the steamy gas emanating from fumaroles in
the crater. We worked with NASA Ames to develop
Dante’s

human
-
machine interfaces, remote science and public interaction capabilities.

Explorat
ion Autonomy

Comprehensive remote exploration requires robots that operate without continuous oversight
and human
-
centered planning. Upcoming planetary missions and terrestrial expeditions call for
robots to traverse thousands of kilometers of uncharted te
rritory under severe environmental,
communication and control constraints. Though short periods of human/robot interaction may
provide mission scientists with enough information to direct robotic exploration of small areas,
supervised science during long,
out
-
of
-
view traversals is impossible. Autonomy will enable
exploration robots to characterize much of what they see and to make the unprecedented
discoveries that will change our knowledge of the universe.

Nomad

perceives and plans routes through rough ter
rain, and employs Bayesian classification
with visual, spectroscopic and metal detector data to distinguish indigenous rocks from
meteorites. The robot also tests various ground search schemes in combination with solar
power resource optimization.
Nomad

is

the first intelligent robot to exhibit advanced cognitive
capabilities for exploring new worlds and to make autonomous decisions about geology and the
origins of inorganic matter.

Dante

Center for Robot Intelligence

Carnegie Mellon University


CONFIDENTIAL

15

Nomad

Pi
on
Pi
onee
r
Pioneer

Skyworker

The
Nomad

robot demonstrated 200
-
km robotic
traverses of a planetary analo
g region, the Atacama
Desert, in 1997 and achieved the first discovery and
in situ
classification of extraterrestrial meteorites in
Antarctica in early 2000.
Nomad

technology
development and both field missions were done in
collaboration with NASA Ames.

R
obot Worksystems

Radiation, extreme temperatures, and other hazards
motivate the use of robot worksystems to do the work of
humans. The sheer magnitude of construction work, followed by
years of inspection and maintenance exceeds what can be
performed aff
ordably by astronauts; the majority will have to be
performed by intelligent robots, sometimes working with people
and at other times operating autonomously.

Robot worksystems like
Pioneer

operate in radioactive
environments like Chernobyl, deploying envir
onmental
characterization sensors, using stereo vision to inspect and
record three
-
dimensional models, and drilling to assess
structural materials. Pioneer combines the dexterity of a robot
manipulator with the forcefulness of a tracked construction
machin
e, giving it a wide range of task capabilities.
Pioneer

was
developed in a collaborative project that included JPL and
Ames.


Skyworker

is a prototype robot for orbital assembly,
inspection and maintenance. A combination of
perceptive sensing, anticipator
y planning and clever
mechanical design allows
Skyworker

to isolate its
payload as it maneuvers across large, delicate space
structures. Higher
-
level autonomy is being developed
for Skyworker and peers to cooperate in assembly
tasks. Our NASA collaborato
rs in the
Skyworker

project
are JSC and Ames.


5.2

Providing Quality People to NASA

Throughout Carnegie Mellon’s history, over 70 alumni
have worked at NASA in a multitude of roles from division chiefs to key scientists, and
astronauts such as Judith Resnik, J
ay Apt, and Edgar Mitchell. These graduates have come
from a variety of fields: robotics, physics, various engineering disciplines, and management.
Recent Carnegie Mellon Robotics Ph.D. graduates now at NASA include Kim Shillcutt at JSC,
Chris Leger and Ma
rk Maimone at JPL, and Justin Boyan and Liam Pedersen at Ames. Our
graduates are emerging as leaders in the fields of robotics, AI, and software engineering within
NASA. Carnegie Mellon is the world’s only university offering advanced degrees in Robotics.

Center for Robot Intelligence

Carnegie Mellon University


CONFIDENTIAL

16

5.3

Carnegie Mellon University Environment

Carnegie Mellon is a pioneer and pre
-
eminent presence in robotics, AI, and software
engineering. CMU’s Herbert Simon received the Nobel Prize in Economics for work related to
his founding role in the AI field, where

CMU remains a world leader. CMU hosts the national
Software Engineering Institute, a renowned Computer Science Department, and offers
specializations ranging from Theory to Human
-
Computer Interaction. The Carnegie Mellon
Robotics Institute is unique in
bringing researchers from the many disciplines related to robotics
under the scope of one organization. These strong interdisciplinary ties are characteristic for
the university in general. Space
-
relevant research has been done in such varied groups as t
he
physics, materials engineering, and cognitive psychology departments.

Carnegie Mellon has pioneered many of the basic ideas broadly used in robotics, including
navigation, vision, planning, autonomy, robot architectures, a wide range of manipulators and

a
variety of mobility modes. Besides conducting basic component research, CMU has integrated
these robotic functions into operational systems that have performed in nuclear environments, in
harsh climates such as Antarctica, and in industrial application
s like agriculture and mining. We
have also applied these technological advancements to robots we are developing for use in
space, and have long
-
standing collaborations with NASA in space robotics research.


Center for Robot Intelligence

Carnegie Mellon University


CONFIDENTIAL

17

6

Organization

The Center for Robot Intelligenc
e is organized to fulfill on its mission to develop both technology
and leaders, and to leverage its relationships with NASA, aerospace companies and its parent
and peer institutions. Further, to better facilitate coordination, cohesiveness and continuity

between the research and educational agendas of the Center, most personnel have
responsibilities to each. The functional organization of the Center is depicted below and
descriptions of the roles of Center personnel are in the sections that follow.

6.1

Manage
ment Team

Director/Chief Scientist

The Director has comprehensive responsibility for the performance of the Center, the quality of
its output, and its relevance to NASA. He expresses the Center’s vision and goals, translates
them into strategy, and defines

and sustains the organizational culture and tempo of work. The
Director has global perspective of intelligent space robotics technology, research and
development. He is the principal liaison to NASA and maintains a perspective of contemporary
and future N
ASA programmatic needs. He is also the principal interface to and is responsible for
development of strategic relationships with the aerospace industry. The Director recruits the
Associate Director, the Education Director, and all Scientist/Educators; he i
s the Center’s
Full Time

Students


Scientists/

Educators

Professional

Educators


Education

Director


Technical

Staff


Director/

Chief Scientist


Associate

Director


Part Time

Students


Advisory

Board


Visiting

Researchers

Education

Research

Center for Robot Intelligence

Carnegie Mellon University


CONFIDENTIAL

18

interface to its Advisory Board; and he is the Chief Scientist of the Center's research program.
50% of the Director’s effort is dedicated to management and marketing of the Center.

Dr. William "Red" Whittaker will be the Center Director and

Chief Scientist. Red is one of the
world's preeminent roboticists and has made numerous contributions to intelligent space
robotics. Throughout nearly 20 years of research and education, Red has demonstrated the
capacity to lead major programs and teams o
f other scientists, technical staff and students.

Associate Director

The Associate Director is responsible for execution of the Center’s strategy, supervision of its
research and education programs, and general operation of the Center. This includes
transl
ating strategic goals into action plans and schedules; allocation and coordination of fiscal,
personnel, and infrastructure resources; budgeting and financial tracking; recruitment, direct
management and review of professional staff; design, development an
d maintenance of
facilities and other infrastructure; marketing; and all reporting, including annual summaries and
research publications. The Associate Director reports to the Director, supervises the Education
Director and Scientist/Educators, is the host

contact for visiting researchers, and interfaces to
other University offices in the context of Center business.

The Associate Director is a senior scientist with a notable career in robotics and proven capacity
to manage research programs and associated i
nfrastructure. He has Faculty standing in
Carnegie Mellon’s Robotics Institute and will be recruited by the Director. The Associate
Director will spend 75% of his time managing the affairs of the Center and the remaining 25% of
his time engaged in research

and/or education.

Advisory Board

A six
-

to nine
-
member Advisory Board assists the Center in maintaining continued relevance to
NASA. It helps the Director define goals, advises him on development of strategy, and provides
additional interfaces to the spac
e robotics research, development and user communities. The
Advisory Board is comprised of leaders in the NASA, aerospace industry and robotics
communities. NASA representatives provide insights to Agency agenda, mission goals,
technology needs, and staffin
g requirements. Industry board members bring perspectives of
current and future needs in the commercial development of space; they add value to the Center
by identifying and helping create strategic collaborative relationships with industry partners.

Throu
gh existing contacts in NASA and the aerospace industry, and with the help of senior
officials at Carnegie Mellon, the Center Director will recruit members of the Advisory Board.
The Board will meet on a quarterly basis to review Center strategy, work pl
an and
accomplishments. While their services are pro bono, Advisory Board members will be
compensated for travel costs.

6.2

Research Team

Chief Scientist

The Center Director spends an additional 25% of his time acting as the Center’s Chief Scientist.
In that

capacity, he is responsible for the quality and integrity of Center research, ensuring that
the research agenda is consistent with the Center mission to deliver intelligent robotics
technology to the space community, and setting the pace and ambition of a
ll research projects.

Center for Robot Intelligence

Carnegie Mellon University


CONFIDENTIAL

19

Scientists

The Center’s research efforts are matrixed by thrusts (planetary and space operations) and
themes (exploration and assembly/inspection/maintenance). Each thrust and each theme is
lead by a scientist. Together these leaders
develop a research strategy that fulfills the Center
goals and complements its educational program. They define and coordinate research agenda;
negotiate allocation of Center personnel and other resources; manage the execution of research
activities; and s
upervise the technical staff. In the context of research efforts, scientists report to
the Associate Director.

Center Scientists are members of the Robotics Institute faculty in its Research Scientist,
Systems Scientist or Professor track. The Center in
tends to 25% support to each of four
scientists (for a total of one full time equivalent) from core funding; the remainder of their
support will be from other sources. Current Robotics Institute faculty already engaged in
intelligent space robotics resear
ch, including Reid Simmons, Sanjiv Singh, David Wettergreen,
and Dimi Apostolopoulos, are the most likely candidates for roles in the Center. Their
biographical sketches are presented below.

Technical Staff

The Center will continue Carnegie Mellon’s trad
ition of making notable achievements in space
robotics technology through projects like Ambler, Dante the Atacama Desert Trek and Robotic
Search for Antarctic Meteorites. Such projects are successful in part because they demonstrate
new research results,
and that aspect demands professional hardware and software engineers
dedicated to the project. The Center will support a technical staff of three full time equivalents.

Full Time Students

In steady state operation, achieved in Year 3, the Center supports s
ixteen full time graduate
students. All but the first year students are connected to one or more research projects. At any
given time, the Center is engaged in several projects, some funded directly through the RETI
Cooperative Agreement and some by contr
acts and grants from other sources. The former are
profiled below.



(1) large project

(2) medium projects

(8) small projects


#

% each

FTEs

#

% each

FTEs

#

% each

FTEs

scientists

3

25%

0.75

1

25%

0.5




staff

3

100%

3.00







students

6

100%

6.00

2

50%

2.00

1

50%

4.00


Large projects are 12 to 24 months in duration, and have at least one major demonstration of
integrated research results. Medium projects are staffed by two students and a scientist who
focus on one or two technologies for 24
-
36 mont
hs. Small projects are the individual thesis
research of graduate students.

6.3

Education Team

Education Director

The Center has a full time Education Director is who defines, develops, and implements
curricula that satisfy both the educational needs of NASA
and the academic requirements of the
Carnegie Mellon. He is responsible for marketing to and recruiting students; tracking and
Center for Robot Intelligence

Carnegie Mellon University


CONFIDENTIAL

20

assessing student performance; and development of distance learning/continuing education
materials and mechanisms. The Education
Director reports to the Associate Director, works with
Scientist/Educators to match students to research, and interfaces to academic officers in the
parent department, school and university.

The Education Director has a doctorate in Computer Science or Rob
otics and prior experience
as an educator at the graduate level. While he will be recruited by the Director, he will also have
to meet the hiring criteria of the Robotics Institute as he will be part of its Teaching Faculty. (i.e.,
the Professor track).

Sc
ientist/Educators

In addition to their research responsibilities, each scientist develops and teaches one or more
courses in the intelligent space robotics curriculum, typically one course per academic year.
The scientists also and advise graduate student
s in the program, guiding them in course
selection and thesis research. Once the Center achieves steady state operation, the student to
faculty ratio is approximately 3.7:1. In the context of education, scientists report to the Education
Director.

Profess
ional Educators

The Center employs one full time equivalent professional educator to design and deliver its off
-
campus education program. Using tools in place at Carnegie Mellon, and the curricula
developed for on
-
campus instruction in intelligent space r
obotics, professional educators will
develop distance learning infrastructure, delivery techniques and courseware. Professionals will
deliver live instruction at NASA facilities and private companies as demand dictates.

Full Time Students

Full time graduat
e students are enrolled in the Carnegie Mellon Robotics Institute (RI) graduate
program as either a Master's or Doctoral candidate. As such, they must not only meet the
Center's criteria for admission, but be admitted into RI's highly selective program as
well. All
students take courses in the fundamentals of robotics that are part of the RI curriculum, core
courses in intelligent space robotics, and round out their course load with electives offered by
the Institute and the Center. Typically, Masters stud
ents matriculate in 3 semesters, while
doctoral students matriculate in 5
-
6 years.

Part Time Students

In addition to curricula geared to full time students, the Center will offer educational products
ranging from seminars and symposia, through certificate
and off
-
campus Masters programs.
Except for those seeking a degree from Carnegie Mellon (who must meet the Robotics
Institute's graduate student eligibility criteria and be selected for admission by the faculty), entry
qualifications for part time students

will be relaxed. Instead, the student's employer will play the
deciding role in determining if the student is qualified to participate, while the Center will
establish guideline criteria that ensure prospective students are well prepared and likely to
suc
ceed.

7


External Relations

Collaboration has always been a vital part of our research and development programs in
intelligent space robotics, as illustrated in the attached figure, and will continue to be
Center for Robot Intelligence

Carnegie Mellon University


CONFIDENTIAL

21

emphasized in the activities of the Center. We have
cultivated deep working relationships
through joint engineering of robots and software, development of science payloads and operator
interfaces, and mission design and execution. NASA scientists have analyzed data and
specimens our robots have collected in

exploration of volcanoes, deserts and ice fields. The
Center's educational program will bring a new dimension to these strong relationships and
foster new ties to NASA Centers. Through the course of prior research and development, we
have also developed i
ncreasingly stronger ties to the commercial aerospace sector; those, too,
will flourish through both Center R&D and education.

A unique opportunity exists in the recently created formal relationship between Carnegie Mellon
and the Ames Research Center. Int
ended to provide NASA an outlet for world class academic
research and education in information technology, the Ames/Carnegie Mellon connection is an
excellent mechanism to infuse intelligent space robotics technologies into NASA.
Center for Robot Intelligence

Carnegie Mellon University


CONFIDENTIAL

22

Prior and current collabor
ations of Carnegie Mellon, NASA and aerospace companies in robot intelligence




Ambler

'89
-
'92


Tessellator

'90
-
'94


Dante I

'92


Dante II

'93
-
'94


Atacama
Desert

Trek

'96
-
'97


Deep
Space
One

'97
-
'98



Tele
-

Robotics

Inter
-

center

Working

Group

'89
-
98



Mars
Autonomy

'98
-
'01


Antarctic
Meteorite
Search

'98
-
'01


Distributed
Intelligent
Robot
Architecture

'99
-
'01


Skyworker

'99
-
'02


Solar

Blade

'99
-
'02


Mars
Sample
Return
study

'01

National

Robotics

Engineer
-
ing

Consortium

'94
-



Hyperion

'00
-
'
02

Ames
















Glenn
















Goddard
















HQ
















JPL
















Johnson
















Kennedy
















Langley
















Boeing
















Lockheed
Martin
















LunaCorp
















MacDonald
Dettwiler
















Metrica
















Orbital
Sciences
















SpaceDev
















TRW
















Intelligent Space Robotics Center

Carnegie Mellon Unive
rsity


CONFIDENTIAL

23

8

Facilities

Consistent with its mission to provide NASA with excellence in intelligent space robotics
research a
nd education, the Center has and maintains a premier physical and computational
infrastructure that supports and facilitates the Center mission. The heart of the Center is a
contiguous set of spaces for prototyping, fabrication, experimentation, and instru
ction, all
located within the buildings that house Carnegie Mellon’s Robotics Institute and School of
Computer Science. This nucleus not only contributes to the Center’s identity as a team
dedicated to creating the robotic technologies for planetary and sp
ace exploration and
extraterrestrial facility development, but more importantly promotes synergistic work through
collocation. Further, its location with the campus provides ample access to and by the entire
Carnegie Mellon robotics and computer science co
mmunity.

The Center’s computational network includes over 100 high performance PCs and workstations
for scientific computation, drafting and mechanical design, robot simulation, communications,
and administration. Further, as part of Carnegie Mellon’s Sch
ool of Computer Science, the
Center has access to one of the best equipped and configured centers of computing in the
world, including the Pittsburgh Supercomputing Center. As a unit within the university, the
Center leverages substantial academic discount
s on industry
-
standard software packages for
mechanical and electronic CAD, analysis, and simulation, real
-
time control system design and
development; and manipulator and mobile robot simulation. The Center maintains several in
-
house libraries of code for
control, communications, user interface development, and image
processing developed through man
-
decades of space and field robotics R&D; it has access to
world
-
class software development techniques and expertise through Carnegie Mellon’s
Software Engineeri
ng Institute.

The Center’s mechatronic development facilities include a CNC milling machine directly linked
to its computing network for art
-
to
-
part mechanical fabrication, as well as conventional
production equipment in a dedicated machine shop. In addit
ion, there is a separate
woodworking and model making shop and two 8000 square foot highbays, equipped with total
coverage cranes, a full complement of tools, and electric, hydraulic, and pneumatic power
sources. The electronics design laboratory features
dedicated CAD workstations, analog and
digital logic analysis equipment, a full array of fundamental prototyping and production tooling,
and an extensive library of current technical literature.