|
Items in cart [0] |
Questions? Call 888-624-8373 |
| Copyright © 2009. National Academy of Sciences. All rights reserved. Terms of Use and Privacy Statement |
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 17
KEYNES ITS:
Ht=N FACTORS RESEARCH FOR THE NASA SPACE STAIlON
Allen Newell
Symposia are held for many reasons. This one is to do a task. I hope
you are intellectually stimulate by what you hear and will take away
some new knowledge that you do not already possess. I also hope the
symposiats have enhanced their own knowledge by.therr studies and are
gratified by the chance to advance their views from this earth-bound,
but otherwise splendid, platform. But neither of these has much to do
with the actual reason for this symposium. We are here to help shape
the research that NASA will perform on how humans interact with the
technology of the space station.
In a nutshell, and to state what I hope is already shared knowledge
among us, operating in space poses extreme challenges. It is a
. . . . . . . . . . .
nostlle, ever elve, constra m m g and untorglv Meg environment. our
intent ~~ a species to make such operations successful and to
continuously extend their scope in complexity, duration, and usefulness
is epitomized in NASA, and other space agencies around the world. It
takes its concrete form by the posing of specific projects, each more
daunting than the last, but (skillfully we hope) set just within the
bourns of the reachable. For us tony bat project is the space
station, a project winch an initial development phase prior to launch of
a decade and a total lifetime of sever more decades. Such projects
force us to not orgy use the best available technology arm science, but
to en them substantially. For us today, ache question is what
research is most needed that card have important payoff for the space
station. It is not possible, of course, to encompass in a single
effort all the technologies and science-= that support the space
station. Thus, we focus on how humans and technology can cooperate to
carry out the operations of the space station.
We will speak today almost entirely of the space station. That is
proper, because we need projects to give as much form as possible to a
future which is almost agonizably open. But, such far-future projects
are emblematic of entire technological futures. Thus, behind the space
station is to be seen an entire spectrum of future space systems,
replete with auto meted and robotic devices, while also being a habitat
for humans. Indeed, the space station is itself not a single
envisioned system, but a projected series that stretches out in time
and evolves in significant ways. Only occasionally will we have need
to distinguish even between such relatively concrete visions as the IOC
17
OCR for page 18
18
and SSOC. The research talked about here in the context of the space
station is what we see as necessary to this entire technological
future.
This is our task. Its success can be measured by the influence of
this symposium on how humans and technology actually work together. Do
the crews that run the space station, both on the ground and in space,
have an easier, safer, more productive time than would otherwise have
happened? We are only one player in the hundreds of individuals,
groups and organizations that affect what goes into the space station,
and a highly transient one at that. Our only leverage is the cogency
of the ideas we put forth. Still, we fail if nothing ~ ~ n stream is
different because of what we say here today. It is not enough to have
an effect, it must be the right kind of effect at the right place and
time. Furthermore, the effect depends not only on NASA decisions about
its research program, but also on the quality of the research that is
thereby enabled, and whether its results transfer into the operational
space station--a notoriously tenuous= conjunctive chain. Still, though
we talk here today of possible research, we hope for operational
results.
My Ok, right now, is to get us launched--to set the stage and
provide the context for the papers you will hear ever the next two
days. I will only take a few minutes to do this. But some overview
will help us to keep on track throughout ache meeting.
1
Hi:
FIGURE 1 Artist's construction of the Space Station, due to launch
1996
OCR for page 19
#
19
THE SPACE SIAITON
Let's start with the space station itself--although I am hardly the one
to do so, with an audience that contains many with active
responsibilities for it. Still, even I know enough to start with the
obligatory picture (Figure 1). This is of course a fantasy, composed
from the minds of many persons and living only there and derivatively
On the minds of receptive audiences. In accordance with its fantasy
character, it changes continually--if not daily, at least monthly. The
planners harden the fantasy with physical mcKkNps that can be walked in
and gawked at. That helps, but the time constants to realization are
still of the order of half-decades.
This wav of talking ascot the scare station mav induce a sense of
, _ .. ,
_ _~ ~ A ~ ~ _ _
fragili by.
That could De a goon thing, if it brings with it an
Increased sense of commitment to making it happen. However', my actual
objective is to induce a sense that much can change in the space
station before it takes its place in the sky and, indeed, after it
does. If we are to consider' launching research in 1987 and expect it
to have cgerational impact, then the time scale of that operational
world must be sufficiently long and its character sufficiently
malleable.
Planning--even research planning--must have same grip on reality.
Thus, we neck to focus on the hard constraints on the space
station ~ the ones that appear to hold no matter what, and on which we
can build securely. Table 1 presents three handfuls--alrea~y more than
can be assimilated in an introduction. mese constraints are what
strike a technically observant human-factors specialist immediately
upon hearing a briefing on the station.
They are the constraints that
shape the roles that humans must play and the tasks they must perform
to make the space station function. What makes them unyielding is the
IABlE 1 The Hard Constraints that Apply to the Space Station
1. Long lifetime of the station (decades).
2. Medium term crew residence on board "months).
3. Small qrcup of residents aloft (less than ten, to begin with).
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
, ,
Large group of operators (non-residents) aground (hundred=).
Very small amounts of r Sources available per resident.
Very ~~a11 amounts of space available per resident.
Infrequent physical communication (months).
Continue us, but limited-bandwidth communication.
Time delay of station communication of .5 to 2 seconds.
Modest time constants of action (minutes to hours).
Weightlessness.
Cont~nuc us, high task load.
Continuous high threat-leveI of many potential errors.
Continuous public exposure.
Completely artificial environment.
OCR for page 20
20
limited state of cur space technology, the primary goals set for the
station, and the necessity of acquiring certain experiences as stepping
stones to future technological frontiers. No matter how technology
changes we must pass this way to move forward--not, of course, with the
exact particularities of the space station we will build, but through
something with the general characteristics listed in Table 1.
Many familiar things follow fort this: the general strangeness of
the weightiness world an] its frustrations; the isolation of the station
grc up, coupled with the lack of privacy and the extent to which members
are locked in; the public work-oriented, regimented world; the complete
dependence on the efforts of others; the stress of bonny living
close to fatal errors. By and large, humans respond adaptively to all
these conditions. Still, humans An space must s ~ their psychic
resources to cope with these conditions, rather than spend it in other
more productive ways.
One striking thing is how saturated with technology the life of the
station will be. This is completely true of those stationed aboard,
but is almost as true of those aground for their workaday world,
although they get to go home to the grass each evening.
Another strik mg thing is that the residence time-scale is long
enough so that many functions have to be ao30=modated that can be
avoided An shorter flights. m e station appears to be a microcosm of
life- so many activities must occur that one can find any problem or
took one looks for, or at 1-ass a close analog. Now, An fact, this is
not quite so. Many function_, such as raising a family, becc=ln;
educated, maying to a new home, and planning retirement, do not show up
at time scale= even as long ~.c months. And to those concerned with the
man-machine ~ m On the modern fighter plane, where the focus is on
action_ 'n the subsecond range, the station will appear downright
leisurely. That the space station occupies a middle range in the total
timescale of human action is a significant simplification--a we will
discover when we have to plan permanent space or lunar stations. But
even so, from the perspective of a human factors analyst, the space
station has moved a long ways toward total living and not just
temporarily occupied workspace. Along with that has come an almost
un-cnw=erable collection of tasks that humans must perform, and the
need for designing the artificial environment in which to perform them.
Still the tasks must be enumerated. One of the great liabilities of
technologist environments is that they don't take care of
theme elves not yet and for same time to come. m e tasks to be
performed in the station and between ground and station most be
enumerated and explicitly planned for. What we fail to enumerate here
below is in parlous state up above. There will always be true stories
about the navel activities of intelligent astronauts, solving
life-critic=] problems or having fun ~ wars we could not Predict.
Bless them for that.
to the need for less preparation. And preparation Impales expects
balk description and enema ration. NASA, of course, has gone to great
lengths to do this. Table 2 provides some reminders of what those
enumerations cover.
, _
But let no one argue back from that blessed fact
_ . e a _ a _ e a a
OCR for page 21
!
21
me left hand cold simply lists Ache various subsystems involved,
so one gets same notion of divers)=. With respect to each of these
there are dry actual tasks to be performed. to enumerate then is to
des ~ ad into the technological gritty of each type of system. But
various types of activities that go into these tasks can be identified,
which is what the right hand column shows. These generic activities
come ~ indefinite variety as well, in terms of what must actually be
accomplished, with what initial knowledge, an] against what
constraints. Finally, ~ have put across the bottom what is perhaps the
most important factor, namely, that the time scale aver which these
tasks endure stretches from less than a second to about four
months -seven powers of ten. Each task in its individu~Ji~y fits into
this time-stretch at some point. But every *oration contains t=.=ks of
every type.
m ere are two points and one conclusion to be made from all this.
First, ~ would impress upon you that there are an almost unimaginable
variety of tasks, which contain almost any combination of task demands
one cares to contemplate. Second, the vast majority of these Inks are
to be accomplishes! by same c~nbinErtion of hens and technology. To be
sure, at the tip ultimately there is a pure human, if only a
congressman; and at the bottom there is a pure machine, if only a
pu~tton making an electric contact. It follows that we can
consider t - my only a selection of all the problems. We will of cue
seek for reseat that is generic In its character and that will impact
large classes of these tasks. But Munich that is important will not ever,
be mer~tioned.
]=CE 2 Types of C>n-Station Tasks for ache Space Station
Hysteric;
Generic functions
Pare':
Guidarx~e & navigation
Fornication & tradki~
Day hailing
P~lsion
Err~rin~nt=1 control and
life sort
Structures/mechanisms
crew systems
Payloads (regiments,
manufacturer, Observations)
Parer handling
Ch~ut
Mechanical ac~tion
Data harrying and ~ = nication
Monitoring/control
Computation, decision and planning
Fault diagnosis and handling
Sensing
.] see 1 see
~ see
10 sec 100 see 103 see
1 min
see
sec
1 hour 1 day
month
OCR for page 22
22
TO TECHNOI=Y OF INTERACTION
me classing situation of human factors has been ~t an irx~ustrial or
mid ivory organization develops some machine to do scam tack. me
h~nan~rator asters of conJcrollin~t this machine are] of being trained
to do so are Sac with in due course. In the best of music, this
occurs early enough to permit modest alteration of the engineering of
the interface. But in the main, the technology of the machine is
autonomous and fixed.
With the advances in artificial intelligence and computer science in
general, and in computer interfaces in particular, the situation is
changing--and changing in several ways simLltanecusly. First, the
machines are becoming more complex, yet capable of more autonomy and
Intel igence at the same time. Second, the interfaces themselves are
becoming more intelligent so that they can aid the user and cooperate
cc~peratively with him. Third, all interface= are becomm g alike in
their utilization of a common hardware and software technolcgy.
Finally--and of a different order entirely--the technolcgy on which all
this is based in itself undergoing rapid evolution, so that all the
features just mentioned are not new fixities that can be depended upon,
but are themselves on the move. All of these current truths have
double for me for the space station, which is located a long ways in the
future. Let ,,~ focus on each of them in turn.
Machines are controllable arrangements of matter and energy that do
things to the physical world. (Thus, toolsaremacolnes.) The ability
to be controlled is of their essence, for it is bat changes then, AL
hero parts of the world, fray a thing that can be taker advantage of
(as to drink flit i a brooklet happened upon) to a thing that can be use
at will (as to turn on a faucet wherever Jchi~s=~. So machines bring
with then the problem of the h~nan-machine ~n~rfa~-, and necessarily
those interfaces are dynamic and continue throughout the duration of
use.
As machines become more capable, through the rational foresight of
their designers and the skill of their builders, the tasks that
machines can do without human intervention increase. Although the real
measure is ~ the iotas range of useful tasks they can accomplish with
acceptable reliability, an appropriate indicator is be l~h of time
machines can go without interaction with humans. WiJch this increased
she cams inevitably the parables of who shed do a Cask, the human
or the machine. Formally, this is exactly the same as the problem of
nether this human or that should do a task, or nether this machine or
blat. Hover, because of the category difference, the htunan-madhine
Cession is in to have a mom proffer character and it becomes the
focus of scientific attention. It is a surrogate, of course, for cur
new to understarx] the advancing capabilities of madhin~c.
m at question is finally abcut to Charge its form radically. me
advances in computers and computation have now been driving
exponentially for forty years. All parts of that advance are
significant for us t~v, in part because they all interrelate. m e
driver of it all, we always say, is the cost/performance of the
computing devices and the level of their integration. But by this time
OCR for page 23
23
that itself depends on software design systems with quality graphics.
So it is all one ball of wax. Nevertheless, the parts here the
advances tough us the most here today is in r~30tics, artificial
intelligence ant Me technology of the h~nan~r~uter interface.
Through these, the account of ~nt=~ligence that can be incorporated into
machines is now reaching the place where the problem of assignment of
functions to men or to machines no Former holds any charm. The
question Test be F ~ s ~ How can h ~ no ark technologies cooperate to
attain a set of syst~m-level goals.
The situation at the interface between the human and some machines
provides a good example of the increase in the capabilities that are
available, with a concomitant increase in the complexity for those of
us who design and understand these systems. As machines increase in
capability, interfacing to them becomes a complex balk ~ its own right
and requires substantial knowledge about what is required to
communicate knowledge back and forth languages, protocols,
communication over intermediate links, the status and location of the
communicants, and on into the night. The solution is to have special
agents that have this knowledge or know how to acquire it, ~ short,
intelligent interface agents. But such agents imply that knowledge
about how things work will be distributed woof what good are such agents
unless they relieve other parts of the system of the responsibility for
having Pertain knowledge and skills? But this reinforces the point
made earlier that it no longer makes any sense to cast the problem of
how humans work with technology in exclusive terms of who controls
wham. Rather, it must be in how agents embodying distributed sources
of knowledge cooperate.
One more point about the technology and I am done with it. If LISA
had to settle for the level of ~nte71igence in current robotic and
expert systems, this symposium would have a very different character.
We have, of course, come a long way in computer science in the last
forty years an] this is plainly evident in existing robotic and
~7 ligent systems. But the changes are proceeding very rapidly and
substantially more capabilities can be expected to be available in
another five years or in five years more age m. This introduce
uncertainty into cur proceedings, for we must not only talk of what new
research Bight bring, but must place this against a background that
will increase in possibilities no matter what NASA does. But this same
motion also adds to the sense of excitement of the new powers that are
possible in the space station. The space station, by being a project
measured in terms of decades, both suffers and benefits in the extreme
from this motion of technology.
RESF~H OBJECTIVES
Given the picture just sketched of Arks and technology, the question
of the day is what research should be done. The substantive answers to
that question are the resE'onsibili~, of the spea~rs of this
sy~imn. ~ would orgy ask you to keep three general considerations
. . ,
On me ..
OCR for page 24
24
First, the research topics raised here range widely from artificial
intelligence, to the human-computer interface, to telerobotics, to
issues of social organization. These are not just a congeries, brought
together to obtain coverage. They are all facets of how humans are to
interact with the primary technology of the space station, and what
technologies are involved in that interaction. A research program
needs to address all these aspects in some coherent way, and not treat
them as separate questions.
Second, we have had to sample to focus on some issues and to
neglect others. But the research program needs to consider the full
range of phenomena. It is in research plans, and the study efforts
that support them, that one engages in the compulsive attempts to
taxonomize the domains and worry seriously about coverage and missing
elements. A symposium is to make clear the fruitfulness of research
areas and to show that there are exciting research questions. Attempts
at completeness and evenhandedness WoN1d only dull the senses. Third,
with more glibness than honesty, I have just shifted an immense burgeon
flus the symposium spearers to the symposium participants ~ or at least
some of them. For, of course, the domain of research is so broad that
coverage is a chimera. blat is especially true if one thinks of
research as devoted to getting answers to specific questions about a
specifically configured space station. Such answers must be
obta~ned~that is what engineering requires. And in the present
context it is human engineering and even organizational engineering. A
research program that is in effect a systematic and planned program of
human and organizational engineering, with the resources to do some
background studies, cannot possibly provide the coverage that is
necessary. Emus, the research program must be aimed at discovering
conceptual, theoretical and technical tools ~ ~~ ~ ~
and organizational engineering of the sna~-
oreat~r efficiency and accuracy.
that Will permit the human
~ _ _~__ station to proceed with
___ _ , _ ~ Only if a research pro gram advances
the theoretical state of Uhe art, including Therein systematic
organizations of data that permit answering a multitude of questions,
will it-serve NEST in the decades it takes to achieve the space
station.
ME INS~O~L CON~r
Thus far, like a good cobbler, I have stuck to my last, discussing the
substantive issues. But it is important to say something about the
institutional context in which the symposium occurs. Blessedly, I need
not take my point of departure frump the current spasm of reflection,
critique an] renewal that has keen the fallout of the Challenger
disaster. For our timescale is too long for that to count as more than
a transient. At least that is true if NASA can continue in its planful
ways, which it shows every sign of doing. Thus, in setting cut the
institutional context I will not taLk about the microstructure of
command and timing that will, in fact, have the lion's share of
responsibility for whether ardor trace of this sy~osium's efforts
survives these too days. Instead, ~ will point to larger entities.
OCR for page 25
25
Let us start with NASA. It is, of course, the primary player. It
is its space station, after all. Its primary view of human factors
considerations has got to be simply as an ingredient to make the space
station better-- a factor of production, in the econc mist 's sense.
That view leads 1nevit~hly to working backwards from specific questions
about the space station to specific studies to answer them. After all,
~ the logic of planful organizations: To get X, set up a plan for
getting X. Furthermore, the cogency of a plan can only be apparent if
it explicitly and recognizably puts down each step, frum what is
available initially to the obtaining of X. This leads to a thoroughly
applied effort and one characterized by short-range goals with tight
loops of justification. Such a logic is certainly appropriate in
part--after all, if NASH doesn't do the studies to deliver the answers
it needs on the ~tty-grit~y of the space station, who else will? But
the timescale of the space station is long enough so that other
attitude= are appropriate as well. NASA can change the available
science enough to make a difference to the space station itself. And
to do that the rP=p~rch must be launched on a broader and freer path,
letting it pick its way among the interesting questions of today to the
different questions of tomorrow. The issue for MESA then is whether it
will rise above the immediate applied questions of human factors to
which the safety and productivity of the astronauts we 1 force
attendance in any event to the faith that major gains for the space
station can be attained f ~ u supporting heroic long-term research.
Each of us has cur own stories of where such long range research by
an institution has made immense differences to the downstream operation
of that institution. Not being a NA5A insider, TO stories of that ilk
do not come from NINA. But even to an outsider it is apparent that
there must be a whole book full of such stories. After all, space
science is an almost new science, even though, as always with science,
it has a whole tangle of historical roots in early rocketry, astronomy,
and more. And space science is practically a creature of NASA, so NA5A
must know all about the gains from bringing a new science along.
Nevertheless, it may be worth recounting briefly one of my own
stories. This is nARpAts creation of the field of artificial
intelligence and expert-systems technology. D~RPA did not start
artificial intelligence, that occurred in the mid 1950s. But only a
few years afterwards, in the early 1960s, D~RPA began its open support
of that part of computer science. It did so in an essentially free
spirit and mixed with the many other things it was also supporting,
such as time sharing, graphics, mLltiprccessors (Illiac IV) and
networking. The support was substantial, but was far freon being the
d~ating ion In the mix of SPA preens. The important At,
from the present point of view, is that DARPA started its support In
1962. By 1972, a mere decade later, ache first expert system had Sun
to urge Derx3~1 and Nycin. By 1982, only one more decade, the
crucial ization of expert systems had begun . Today, five years
later, though still a green and nascent technology, it has benzene the
property of us all. It has beck ntegral to Ash of DOD's own future
and is near integral to our discussions here. But for a~st all the
first twenty years, PUPA was essentially the only support for
OCR for page 26
26
artificial intelligence. (My friends from ONR and from NIH-supported
AID (AI in medic me) may be a little annoyed at so sweeping a claim;
yet it remains close to true.) Thus, we see that an agency can bring
into existence wholly new techniques and ideal for its own use
downstream. It cannot usually be done In less than a decade. But in
tim£scales that are cam=ensurate with the space station, such things
are possible. And their payoff is incalculable.
The second major player is the collection of scientists and
engineers who will conduct the research. This is not a homcgenous
grc up. post immediately, the scientific cadres within NOVA concerned
with human factors and artificial intelligence are to be distinguished
f ~~ the scientists in the universities and research organizations
across the country. Each clearly plays a different role, although, in
the style of the times, strong attempts exist to weld these into a more
continuous community, with the e=~hli~nt of places such ~~ the NASA
R£ search Institute at Stanford University.
The more important inhomogeneity is among the social institutions we
calf professions an] disciplines. Focus narrowly on the human-science
issue= concern mg the space station, and ignore totally the
half-hundred natural-science and engineering disciplines concerned with
the physical structures in the space effort. However, a gaggle of
disciplines are still gathered around this s ~ ly narrow focus.
Alphabetically, they are: artificial intelligence, cognitive
psychology, computer science, human factors, industrial engineering,
organization theory, robotics, social psychology, sociology. I have no
dcNbt Overlooked some, but all these, at least, are represented among
the speakers of this symposium. The inhomogeneity here arises from two
sources. First, the issues of the space station involve multiple
technologies, and the relevant human Phenomena are so diverse that they
necessarily make contact with different human sciences. But record,
multiple h~nan-science disciplines focus on the same Phenomena, but do
so from different Natives. In particular, the emergence of the
cc ~ uter as a mass ~ibenomena has raised the prcblen of human-comput~r
interaction to prominence. At least four disciplines artificial
1nt=1ligence, cognitive psychology, computer science (mostly graphics
and interface programs m g) and human factored are currently engaged in
forming an inter~iscipl~ne called human-computer interaction (HCI).
The effort is currently focused on the individual in interaction with
the computer via a system of interaction mechanisms (displays,
keyboards, pointers, etc.). It is acknowledging, though only
gradually, social and communicative dimensions. The conceptual and
disciplinary turbulence involved In all this is both part of the
inhomogeneity of the current scene and revelatory of it. HCI is only
one part of the human-relet ~ issue of the space station, though a
significant one.
m e NASA situation that we discuss at this symposium provides an
cpportunity for these disciplines. They can, of course, treat the NA5A
problems as if they were just another collection of interesting
situations in which to ply their investigatory trade. Our
nation blessedly, once aga~n--is extraordinarily pluralistic. Thil=,
NASA research contracts and grants can be taken as providing additional
OCR for page 27
27
m~cro-research opportunities in a larger mix.
important one.
But the NOSE situation provides a larger cpporbunity, or at least it
does if MESA chooses to mate that opportunity available. The space
station provides a unique focus for the development of the science of
how humans interact in a technology-saturated environment. By reaching
so far ahead of the degree of saturation in the rest of current
society, it offers a chance to study a world well ahead of its time.
It is a unique cpporbunity in this historical moment, although it will
become less so as the saturation of the rest of the world proceeds.
It is important to realize that In applied sciences technological
foci have an immense influence on the character of the science. One
Hal only to think of the influence on human factors of its being
nurtured by the aircraft industry, while being relatively ignored by
other industries. mans, NASA has a fleeting opportunity to bend the
twig of HCI to a long-term concentration on aspects especially relevant
to N~SA's interests.
m is is one view and an
ENVOI
the ingredients of the symposium have now been assembled before your
very eyes--the space station; the tacks of human-technology
interaction; the technologies that are both the object of that
interaction and the means to make it work; the orientation towards the
research that needs to be done; and the institutional setting within
which this symposium must make its contribution. Let us now move to
the substantive papers.
OCR for page 28
Representative terms from entire chapter:
artificial intelligence
