systems autonomy - nasa · tcs tcs/power hierarchical multiple sys distributed multiple sys 87...
Post on 17-Apr-2020
3 Views
Preview:
TRANSCRIPT
r-_,
SYSTEMS AUTONOMY
Henry Lum, Jr.Chief, Information Sciences Division
NASA Ames Research Center
t,od_ TECHNOLOGY FOR FUTURE NASA MISSIONS
t
AN AIAA/OAST CONFERENCE
ON CSTI AND PATHFINDER
12-13 SEPTEMBER, 1988
WASHINGTON D.C.
(
\
,2Oo_D
I
.Q
.Q
https://ntrs.nasa.gov/search.jsp?R=19890002402 2020-04-19T05:11:25+00:00Z
SYSTEMS AUTONOMY PROGRAM
MARS EXPLORATION
ENABLING TECHNOLOGIES FOR THENATIONAL SPACE CHALLENGES
PERMANENT PRESENCE IN SPACE
CO
LUNAR OUTPOST• LOWERS MISSION OPERATIONS COSTS
• INCREASES PRODUCTIVITY
• RENDERS HIGHER QUALITY DECISIONS
• MAINTAINS TECHNOLOGICAL LEADERSHIP
TRANSPORTATION
"11
_0 I_
bJ
LD
OO-_:;0
END-TO-END SYSTEMS INTEGRATION OF HUMANS, INTELLIGENT SYSTEMS, AND FACILITIES
25O OF POOR QUALiTf
t",O
Crl
N/3SAAmes Research Center Information Sciences Division
SYSTEMS AUTONOMY PROGRAMWHY INTELLIGENT AUTONOMOUS SYSTEMS
REDUCE MISSION OPERATIONS COSTS
• AUTOMATE LABOR INTENSIVE OPERATIONS
INCREASE MISSION PRODUCTIVITY
• AUTOMATE ROUTINE ONBOARD HOUSEKEEPING FUNCTIONS
INCREASE MISSION SUCCESS PROBABILITY
• AUTOMATE REAL-TIME CONTINGENCY REPLANNING
HL/AIAA 9-88 (LAH)
['-3Cm
N/ AAmes Research Center Information Sciences Division
DESCRIPTION OF INTELLIGENT AUTONOMOUS SYSTEMS
CHARACTERISTICS
KNOWLEDGE-BASED SYSTEMS
• DYNAMIC WORLD KNOWLEDGE ACQUISITION, UNDERSTANDING,
AND EXECUTION OF COMMAND FUNCTIONS
• RELIABLE DECISIONS IN UNCERTAIN ENVIRONMENTS
• LEARNING ABILITY
• ALLOWS "GRACEFUL" RETURN TO HUMAN CONTROL
CAPABILITIES
GOAL-DRIVEN BEHAVIOR
• COMMUNICATE AT HIGH LEVELS WITH HUMANS AND OTHER MACHINES
"COLLABORATIVE" HUMAN-MACHINE INTERACTIONS
• RECOGNIZE AND RESOLVE COMMAND ERRORS
SELF-MAINTENANCE
• OPERATE AUTONOMOUSLY FOR EXTENDED PERIODS OF TIME
!
HL/AIAA 9-88 (LAH)
SYSTEMS AUTONOMY PROGRAM
HOW DO WE GET THERE - PROGRAM ELEMENTS
[,_O7Co
ONGOING CORE TECHNOLOGY
• PLANNING AND REASONING
• OPERATOR INTERFACE
• SYSTEMS ARCHITECTURE
[ i[ fp,_s_,c, 1 ! z_c.,o,o_
,,ooucTs_I i pcus!
PERIODIC DEMONSTRATIONS
• LONG TERM EVOLVING
TESTBED .........
• SHORT TERM SPECIFIC
DOMAIN DEMOS
;-Ii
_k ¸
SPACE STATION
SHUTTLE
MISSION CONTROL
SHUTTLE
LAUNCH DIAGNOSTICS
DO
(J14_
NASAAmes Research Center
SYSTEMS AUTONOMY PROGRAM
TECHNICAL CHALLENGES
Information Sciences Division
REAL-TIME KNOWLEDGE-BASED SYSTEMS
DYNAMI(_ KNOWLEDGE ACQUISITION AND UNDERSTANDING
ROBUST PLANNING AND REASONING
COOPERATING KNOWLEDGE-BASED SYSTEMS
VALIDATION METHODOLOGIES
HL/AIAA 9-88 (LAH)
N/ AAmes Research Center
SYSTEMS AUTONOMY PROGRAM
Information Sciences Divison
- TECHNOLOGICAL CHALLENGES
WHERE WE ARE TODAY
Crl
O1
REAL-TIME KNOWLEDGE-BASED SYSTEMS• NO PARALLEL SYMBOLIC-NUMERIC PROCESSORS
• SLOW SPECIAL-PURPOSE HARDWARE (1 GBYTE MEM, 5 MIPS)
• PROTOTYPING S/W SHELLS (ART, KEE, KNOWLEDGECRAFr)• DIAGNOSIS AND PLANNING DECISIONS IN 1-10 MINUTES
DYNAMIC KNOWLEDGE-ACQUISITION & UNDERSTANDING• NO AUTOMATED EXPANSION OF K-B
• SMALL STATIC PRE-PROGRAMMED K-B
• DEC "XCON" LARGEST (5000 RULES, 2000 COMPONENTS)
ROBUST PLANNING AND REASONING
• HEURISTIC RULES ONLY, NO CAUSAL MODELS
• PRE-MISSION PLANNING (NO REAL-TIME REPLANNING)• DIAGNOSIS OF ONLY ANTICIPATED SINGLE FAULTS
• "FRAGILE" NARROW DOMAINS (RAPID BREAKDOWN AT K-B LIMITS)
COOPERATING KNOWLEDGE.BASED SYSTEMS
• SINGLE STANDALONE DOMAIN SPECIFIC SYSTEMS
• HUMAN INTERACTION ONLY, NO INTELLIGENT SYSTEMS INTERACTION
VALIDATION METHODOLOGIES
• CONVENTIONAL TECHNIQUES FOR ALGORITHMIC SYSTEMS
HL/AIAA 9-88 (LAH)
C_
OPERATOR
OBSERVABLES AND EXECUTION STATUS
1DISPLAYS
INTERFACE
I CONTROLS I
SIMULATOR 1
KNOWLEDGE
MONITOR
ADVISE BASE
OPERATOR TASK PLANNING
EXECUTOR
& REASONING
I DIAGNOSER]
INTERROGATIONS
PLANNER I
SENSING &
PERCEPTION
tINTERNAL
OBSE RVABLES
iCONTROL
EXECUTION
EXTERNAL
OBSERVABLES
STATECHANGES
EXECUTION COMMANDS
SYSTEM ARCHITECTURE & INTEGRATION
KNOWLEDGE
SYSTEMS
b_
Cn
EXTENSIONS, TESTS,
VERIFICATION VALIDATIONS
EXPERTISE
DYNAMICWORLD
"u'8'LEM PEIRCf.V I
J
INTER-
ACTION
KNOWLEDGEACQUISITIONSUBSYSTEM
KNOWLEDGEBASE
EXPLANATIONSUBSYSTEM
REASONINGENGINE
-I)
v
USERINTERFACE
WORKINGq
MEt,'K)RY
AUTOMATED SYSTEMS FOR IN-FLIGHT MISSION OPERATIONS
EVOLUTION OF AUTOMATION TECHNOLOGY
CO
.i-
OO
0_,
_C)'O
r'm
NASA AMES RESEARCH CENTER
OAST-SPONSORED FIEBIEAIRCH
t_C/1¢.O
E/S
TECHNOLOGY
OO
•--I ...
BEFORE NOW
SYSTEMS AUTONOMY PROGRAM DEMONSTRATION
SYSTEMS AUTONOMY DEMONSTRATION PROJECT (SADP)
Oh
0
O_
r" v, _
!::7̧
2227
,"_, y'
PARTICIPANTS AND FACILITIES
PARTICIPANTS
• AMES RESEARCH CENTER• JOHNSON SPACE CENTER• LEWIS RESEARCH CENTER• MARSHALLSPACEFLIGHT CENTER•INDUSTRY
FACILITIES
• ARC INTELLIGENT SYSTEMS LABORATORY• JSC INTELLIGENT SYSTEMS LABORATORY• JSC THERMAL TEST BED• LeRC POWER TEST BED
OBJECTIVES
DEMONSTRATE TECHNOLOGY FEASIBILITY OF INTELLIGENTAUTONOMOUS SYSTEMS FOR SPACE STATION THROUGH TESTBEDDEMONSTRATIONS
• 1988: SINGLE SUBSYSTEM (THERMAL)
• 1990: TWO COOPERATING SUBSYSTEMS (THERMAL/POWER)
• 1993: HIERARCHICAL CONTROL OF SEVERAL SUBSYSTEMS
• 1996: DISTRIBUTED CONTROL OF MULTIPLE SUBSYSTEMS
TCS
TCS/Power
Hierarchical
Multiple Sys
Distributed
Multiple Sys
87
SCHEDULE
88 I, 89 90 I 91 92 93
94 95. 96
HL/AIAA 9-88 (LAH)
1988 DEMONSTRATIONSYSTEMS AUTONOMY DEMONSTRATION PROJECT
SPACE STATION THERMAL CONTROL SYSTEM (TEXSYS)
Oh
¢
.-,,-_,__
qa
- . _[a6,:'.:_.___ _' -1
OBJECTIVES
IMPLEMENTATION OF AI TECHNOLOGY INTO THE REAL-TIMEDYNAMIC ENVIRONMENT OF A COMPLEX ELECTRICAL-MECHANICALSPACE STATION SYSTEM- THE THERMAL CONTROL SYSTEM.
• REAL-TIME CONTROL
• FAULT DIAGNOSIS AND CORRECTION
• TREND ANALYSIS FOR INCIPIENT FAILURE PREVENTION
• INTELLIGENT HUMAN INTERFACE
• CAUSAL MODELLING
• VALIDATION TECHNIQUES
PARTICIPANTS AND FACILITIES
PAR TICIPA N TS
• AMES RESEARCH CENTER
• JOHNSON SPACE CENTER
• INDUSTRY: LEMSCO, ROCKWELL INTERNATIONAL,GEE)CONTROL SYSTEMS, STERLING SOFTWARE
FA CIL I TIES
• ARC INTELLIGENT SYSTEMS LABORATORY
• JSC INTELLIGENT SYSTEMS LABORATORY
• JSC THERMAL TEST BED
Development
Requirements Definition
Design Definition
Integration V & V
TCS Demonstration
Power System Interfaces
TCS/Power Demonstration
Analysis, Reporting
87 88
r_
'--1
=m
SCHEDULE
89 90 91 92
I
!
--=
-'--1
HI.JAIAA 9 88 (LAH)
O_t'O
O0
m
OzO__t-"
_O"u
t-'l-t]
/
t_
C_
NASAAmes Research Center Information Sciences Division
SYSTEM AUTONOMY DEMONSTRATION PROJECTTCS FUNCTIONAL CAPABILITIES
PROTOTYPE OBJECTIVES
CAUSAL MODELS/SIMULATION
LIMITED FAULT DIAGNOSIS
DEMONSTRATION OBJECTIVES
NOMINAL REAL-TIME CONTROL
FAULT DIAGNOSIS AND CORRECTION
TREND ANALYSIS
DEMO
1/87
KNOWLEDGE BASE EXPANSION
1
6/87
©
©INTELLIGENT INTERFACE
DESIGN ASSISTANCE O
TRAINING ASSISTANCE O
2
9/87
3
12/87
©
©
©
®
©
©
4
2/88
®
5
5/88
HL/AIAA 9-88 (LAH)
N/ AAmes Research Center Information Sciences Division
03
SYSTEMS AUTONOMY PROGRAM - TECHNOLOGICAL CHALLENGES
B. WHERE WE NEED TO GO
REAL-TIME KNOWLEDGE-BASED SYSTEMS
• PARALLEL SYMBOLIC-NUMERIC PROCESSORS (100 GBYTES, 500 MIPS)
• NEURAL NETWORKS (BRAIN CELL EMULATION)• LAYERED TRANSPARENT SW
• DIAGNOSIS AND PLANNING IN MILLISECONDS
DYNAMIC KNOWLEDGE ACQUISITION & UNDERSTANDING
• AUTOMATED K-B EXPANSION IN REAL-TIME (LEARNING)
• LARGE DYNAMIC DISTRIBUTED K-B
ROBUST PLANNING AND REASONING• COMBINED HEURISTIC RULES AND CAUSAL MODELS
• REAL-TIME CONTINGENCY REPLANNING
• DIAGNOSIS OF UNANTICIPATED FAULTS
• SPECIFIC DOMAINS ON BROAD GENERIC K-B (GRACEFUL DEGRADATION)
COOPERATING KNOWLEDGE-BASED SYSTEMS
• HIERARCHICAL AND DISTRIBUTED SYSTEMS
• HUMAN AND INTELLIGENT SYSTEMS INTERACTION
VALIDATION METHODOLOGIES• METHODOLOGY FOR EVALUATING DECISION QUALITY
• FORMAL THEORETICAL FOUNDATION
HL/AIAA 9-88 (LAH)
rchitecture of an Autonomous Intelligent System 1
C_
Operator
Displays
Conr.rols
iiiiiiii:i!iiii!iiiiliiiii_iiiiiiiiiiiiiiiiii!i!iiiiii_ii!i!i_ii!i_iii!iiiiiii:iiiii
ii:!!ii
Nominal i
i!Plan!!!il
. • .•.
.... Exr
Simulator : -
Nominal World :::;: Direct World
l¢ : - : :: . :
KNOWLEDGE BASE
• Dynamic World Model
• CAD/C/S_t Data Base
• System Configuration
• Heuristic Rules
:::;; :::::::::
Perceptor
Vernier :i:i:::::: :::::::l;i!ii!ii_i?i!::i!!!_l:::::::::: :::::::
_ntrol i:i:i:i: .!:!:i:i..-=iiiii:.;_;!;:::Internal ::::::lii!i!iiiii!::ii!i!::ili:i::::::1: ::: :!:i:ii!i::_::i:-:i::ii::i31ii::iObservables :::::!il [_:i_i_;i_iiiil;! • . • . - . : • : •
_i!:il il;i!i!i!_!_! ;:[:::;::[:
Effector
. i!Description " _i_i:i:iiiiiiiii . ... :: .... -:::-:.iii I i.:. Nominal Plan!; ...... :
.......... . , . , • . . . . .. .... -..
::: ::!i System Architecture d Integration ; :i::::::.:ii
Task Planning
Sensing &
Percep _l on
External
Observables
State
Changes
Manipulators
Control
hlecha_ iza t/on
N/ SAAmes Research Center Information Sciences Division
SYSTEMS AUTONOMYDEMONSTRATION PROJECT
Technoloqy Demonstration - Evolutionary Sequence
1988
Automated Control Of Single Subsystem
("Intelligent Aide")
Thermal Control System
* Monitor/real-time control of a single subsystem
* Goal and causal explanation displays
* Rule-based simulation
* Fault recognition/warning/limited diagnosis
* Resource management* Reasoning assuming standard procedures
1993
Hierarchical Control of
Multiple Subsystems
("Intelligent Assistant")
1990
Automated Control of Multiple
Subsystems ("Intelligent Apprentice")
Thermal Control System and Power
System
* Coordinated control of multiple subsystems
* Operator aids for unanticipated failures* Model-based simulation
* Fault diagnosis for anticipated failures
* Real-time planning/replanning
* Reasoning about nonstandard procedures
1996
Distributed Control Of
Multiple Subsystems
("Intelligent Associate")
* Multiple subsystem control: ground and space
* Task-oriented dialogue & human errortolerance
* Fault recovery from unanticipated failures
* Planning under uncertainty
* Reasoning about emergency procedures
* Autonomous cooperative controllers
* Goal-driven natural language interface
* Fault prediction and trend analysis
* Automated real-time planning/replanning
* Reasoning/learning, supervision of on-board
systems
HIJAIAA 9-88 (LAH)
AUTONOMOUS SYSTEMS FOR ADVANCED LAUNCH SYSTEMS (ALS)UNMANNED LAUNCH VEHICLES
_3
J
NASA AMES RESEARCH CENTER
OAST/AF-SPONSORED RESEARCH
OhCO
fU/ AAmes Research Center Information Sciences Division
AI Research Issues
• MACHINE LEARNING
• COOPERATING KNOWLEDGE-BASED SYSTEMS
REAL-TIME ADVANCED PLANNING
METHODOLOGIES
AND SCHEDULING
• MANAGEMENT OF UNCERTAINTY
• AUTOMATED DESIGN KNOWLEDGE CAPTURE
• VALIDATION OF KNOWLEDE-BASED SYSTEMS
HL/AIAA 9-88 (LAH)
FOcr__JD
BAD
BETTER
GOOD
BEFORE
MACHINE LEARNING
PREDICTIONS: REMEMBER SEARCH MISTAKES
© O
L I
I O =
I
MODEL REFINEMENT
AFTER
(
,.c7
// \i \
\
( (
Technician
SCHEDULING HEURISTICS
Technicians are in great demand
Histogram Pane
OO"n:_
O_
On_C__0r-r_lR
-'1..
COOPERATIVE INTELLIGENT SYSTEMS
"-3O
0 (3,
F _
C: ',:'='
i,=,, ;,_,
,,,j
DESIGN KNOWLEDGE LOST WHEN DESIGNER LEAVES:
BUT WHY
IS THIS AP
2.7 mm?
CONSERVATION OF DESIGN KNOWLEDGE
WHY DID THEY CHOOSE
SILVER INSTEAD OF STEEL?
ELECTRONIC NOTEBOOK
"(DRIVEN-BY
$OBJECT
MOTOR -3977)"
i KnowledgeIntensive I
• Strong prior theory
• One (or few) examples
• Verification by proof
• Learned concept must
be useful
/
Markov Models
Machine Learning
Knowledge Weak I
• Weak prior model
• Many examples required
• Cannot prove theory
• Learned concept reflects
intrinsic structure
Di_K,_el Discovery_ ?
I Classification Models I
f.- -,,,I Supervised ! i Unsupervised
Series Prediction I
rl'OCl./ Sc
,,ii,__i++
tol
6. 7+ O, _. tO. I1. 12+ 13. 14 15. l&. 17. IO. 1_. _O. 21 • n n _.
WAVELENGTH. _lt
lEE&4.
| ,,.-Zqll 4e
1_3Z
zI_X Iq57q +
i 9716
t,
. ,
& ? I IJ 11 ||. 12 I) 14 I_ 16 17 I| Iq _51 Zl 22. n
WAVELENGTH. _m
_IO,l
_r 45+i
, ° •
41.0+
u
:_ 11 "
+_7 5[ fill
+: 0
• • o • ! _.. " . o• • "oO. •b• • • .o •
, • • .& .... ; _ ;,,b • o,.• _; .. • 0 •................. _ ........ "....,.L l•db-..q_ _.,.-_. _ ,+...++++_.+.+r+..¶.-- _. _ •
• • • * i me ++ ; 00. +e • e•
._+. o .o
_l+ll,
# .!
451
0 2Z
_- • • • • . • ° -" • •
• • • -- +. • • _ • . -- .
r-
-2Z + •
0
• -IV
4hi
-500++5, 45. 51 1]5. llO - IN -1_15 -110 -_5 Ill _$ _IO I_ Ill
GALACTIC LONGITUDI_. _q GALACTIC LONGITUDE, dell
The spectra show two closely related IRAS classes with peaks at g.7 and 10.0 microns.
This discrimination was achieved by considering all channels of each spectrum• AutoClass
currently has no model of spectral continuity. The same results would be found if the
channels were randomly reordered.
The galactic location data, not used in the classification, tends to confirm that the clas-
sification represents real differences in the sources•
NASAAmes Research Center Information Sciences Division
Function
Technology I
Status I
ITechnology I
[ F°recast I
Examples I
Evolution of Advanced Architecturesfor
Real-time, On-board Teraflop Systems
PhotonicProcessors
Coarse-Grained
Parallel Systems
Fine-GrainedArchitectures
RT Image Processing
Applied R&D
Knowledge Under-standing & Control
Development
Deep Reasoning
Basic Research
Late 1990s Current Early 2000s
KBS-controlledPhotonlc Processor
SVMS(6-Processor System)
Neural NetworksFuzzy Logic Computes
and Controllers
|HL/AIAA 9-88 (LAH)
Ames Research Center Information Sciences Division
O1
Computer Architecture Research Issues(Numeric/Symbolic Multiprocessor Systems)
• OPERATING SYSTEMS FOR REAL-TIME MULTIPROCESSING SYSTEMS
IN A HETEROGENEOUS ENVIRONMENT
• VALIDATED COMPILERS AND TRANSLATORS FOR AN ADA-BASED
MULTIPROCESSING ENVIRONMENT
• DATABASE MANAGEMENT FOR LARGE DISTRIBUTED DATABASESGREATER THAN 10GB
• AUTOMATED LOAD SCHEDULING FOR MULTIPROCESSORS
• REAL-TIME FAULT TOLERANCE AND RECONFIGURATION
• RADIATION HARDNESS WITH MINIMUM PERFORMANCE COMPROMISES• PROCESS TECHNOLOGY
• VLSI/VHSIC TRADEOFFS
• EFFICIENT COMPILERS AND INSTRUCTION SET ARCHITECTURES
HL/AIAA 9-88 (LAH)
SPACEBORNE VHSIC MULIIPI4OCESSL)I4 SYSI i-M
NASA/AF/DARPA COLLABORATION
f
PROCESS
VHSIC TECHNOLOGY
0.5H TARGET
1.25_ BACKUP
RAD-HARD CMOS
105 RADS RADIATION RESISTANCE
NO SINGLE EVENT UPSETS
"-4ob
SYSTEM CHARACTERISTICS
• PARALLEL ARCHITECTURE
- 40-BIT SYMBOLIC PROCESSORS
- 32-BIT NUMERIC PROCESSORS
• FAULT-TOLERANCE/AUTOMATED
RECONFIGURATION
• OPTICAL INTERCONNECTS
• 25 MIPS SUSTAINED UNIPROCESSOR
PERFORMANCE (40 MIPS TARGET)
-MINIMUM OF 100 MIPS OVERALL
SYSTEM PERFORMANCE
• DBMS FOR 10G BYTE MEMORY
MANAGEMENT
f
/
/
MEMORY 32 b=t DATA + 8 bit TAG
PROCESSING UNIT
BUS CONTROLLER
MICRO MICROCODE ADDRESSSEQUENCER
GENERATION
MEMORY INTERFACE
S
/
° I'" ° h", J
i#!#" !#'!#" !# i_ i_5-" ,
,.,,.,-,,i,.,,,:.: ,,.,,,. .. ,." ." !.,
._ o . ,........_,i:,_i_i,ii_,i:i:_:i,,,.......
°,, . . _11!
(SVIVI_)
POTENTIAL SPACE&
AERONAUTICS APPLICATIONS
r _ _ .
PHOTONIC PROCESSOR FOR REAL-TIME IMAGE UNDERSTANDING
-,.,j
OBJECTIVES
• REAL-TIME PHOTONIC PROCESSORS& TECHNIQUES
for Terrain Analysis Tasks
• SYSTEM CONTROL & INTEGRATION OFEMBEDDED PHOTONIC PROCESSORS
with Integrated Numeric/SymbolicMultiprocessor Systems
• TECHNOLOGY FEASIBILITY DEMONSTRATIONS
Focusedon PlanetaryRovers & Space Vehicles
BENEFITS
• Real-time, High Performance Parallel Processingfor Image Processing& Understanding
• Fault Tolerance
• Low Power, Weight, and Size
i;',,=_ .
l
!
POTENTIAL APPLICATIONS !
Autonomous Landing
m
Sample Acquisition and Analysis
Sample Return
NASA AMES RESEARCH CENTER
OAST-SPONSORED RESEARCH
0 ::;
OC"Irl _
'_Z ,"
C_ i:-_
L_
Knowledge-Based Systems
The tasks involved with an image-understanding-systemcan be divided into three layers as shown. The problem is
to find a synergistic balance between all layers so that asknowledge of the image accrues, the reliability of the
interpretation, recognition, and enhancement increases,while the amount of required computation decreases.
Methodologies of organizing a knowledge-base of objectand using a rule-based system to effectively search the
knowledge-base and directing the computations of pho-tonic processors are being developed. The majority o1[the
domain specific knowledge for a task will reside in theinterpretative level making the photonic processor ageneral purpose computing tool
ORIGI_AL F::',:I/,.I ,5OF POOR QUALITY
279
b,J
E:)
0
NBSAAmes Research Center
COLLABORATIVE
[ ACADEMIA I
CARNEGIE-MELLON i_
l::_!!_i!i UN VERSITY ::_-il
M ne ,.g ..the,,y I_ Mason
Cooperating Knowledge-
B_edsystems,O=a _:::::-_¥_6i_6_i_:"i::::::::::i_Accelerators,A,/ lii:;iii:i!_i! Flynn i!ii!_ii!ilil
Robotics Intogr_tion, I'::iiiiii Cannon ::::ii_iil"_
and R/T C,o_rol ol !ii_!_iiii: Feigenbaum _;i_i_
TeleautonornYMachineLearningand _::i::i::i::i::i:: Laird i;i!iiiiiii_
Processors ::::::i:::::: Zade_ ::!::!i?:::i::::ii
Man-Machine Inter- _,ii!i!iiiii!iiiiil M I T i ! _,ace, Intelligent On- Eii!iii!iiiiiil;iiii! Young
Space Sciences _!iii_!_!_!_i; SheridanSyste m s =:=====================================::==========
L_, i_?_:_ ,s, _i!i!Partial I_ii!ii::! Rc_enb,oom i::i_ill _Architectures ;:_:;:;:: Mizel ::i::::::
F SPACE TELESCOPE _ F- J L
C DARP^ ) _sMou ,NS_TUTMouE J L Atkins°nGunter
Intelligent Systems KB Systems Ground Data
Research and For Space Syslems Auto-
Advanced Computer Science malion and FlightArchitectures Processors
AI AND COMPUTER
IN VES TIGA TOR
AMES
SHAFTO
CHEESEMAN, BERENJI
FRIEDLAND, RICHARDSON,LESH
ROSENTHAL, DRUMMOND,ZWEBEN, FRIEDLAND
ERICKSON, RUDOKAS
SIMS, FRIEDLAND,
LANGLEY, MINTON,
CHEESEMAN
LUM
SIVARD, ZWEBEN
AI Tech for Shuttle Design Data
and Space Station Caplure withAutomation ST Focus
I GOVERNMENT
ARCHITECTURES
AREA OFINVES TIGA TION
RESEARCH TEAM
[ INDUSTRY I
_il;iiiiiiiiiiiiiiii"Rl_siii_i!i!i_i_i!i_ii::i::_
Operator Interfaces
Reasoning underUncertainty
Validation Methodologies
Planning and Scheduling
Causal Modeling andReasoning
Automated Learning
Symbolic Architectures/
Programming Environment
Automated Design
Knowledge Acquisition
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
ii!i!i!iii!i!i!iiiii.:= :,:!i!i!iiiiiiiiiili
!_i_i_i!i!i_i_i_i_i_i_i_i_i_i_DECi!!iiiiiiiiiiiiiiii
)
Architectures
SAN for Work
Stations
Space Station
Automation
Technologies
Autonomous
Systems
Technologies
Procedural
Reasoning
Beta Test-
Site for Symbolic/
Numeric Processors
and Collaborative
Research
HL/AIAA 9-88 (LAH)
• _ Space Station Automation
_,_ _ (Power)
POCC and K-B Launch Operations Collaboniion between
Systems Distributed Talembotics and
Data Bases Systems Autonomy
Expert Systems
Developmen(
Tools
281
top related