Copyright © IFAC Analysis.

[)csig-n
Copvright Design and
E\-aiu3tion of \lan-\lachine-
E\'aluation \lan-\lachine Systems
Svstt"ms
Badcn·Badcn.
Bacien· BadcIl. Federal Republic of German\' ~.'-I~
Cermany 1qH2

MAN--MACHINE SYSTEMS INTRODUCTION AND

BACKGROUND

G. Johannsen
Research Institute for Human Engineering (FGANIFAT), Komgstrasse,
Konigstrasse,
Wachtberg- Werthhoven, Federal Republic of Germany

Abstract. This paper is an introduction to the IFAC/IFIP/IFORS/IEA Confer-

ence on Analysis, Design, and Evaluation of Man-Machine
Han-Xachine Systems. It serves as
an umbrella for the survey papers and topic areas of the conference. There-
fore, it is very broad in its scope and condensed in its exposition. The man-
machine system is defined, its general purpose explained, and the multitude
of application areas stated. The historical and scientific background of the
field is briefly outlined. Human task categories in man-machine systems are
described.

Keywords. Man-machine systems; manual control; optimal control model; super-

visory control; operations research methodologies; problem solving; human-
computer interaction; software ergonomics; human reliability; social effects
of automation.

DEFINITION AND PURPOSE OF MAN- Environment

SYSTEMS
MACHINE SYSTEHS ,...---
---
r--- - - - --
--- -- - --------,
---- --------- - ------,
,,
I
I Man - Machine System
A man-machine system is defined as a func- ,
I
I
I
I

tional synthesis between a biological/psycho- I

I
I
logical/social system (the man or a group of Goals: I Output
people) and a technological system (the ma- Man Machine
chine) characterized predominantly by the
14====1
14=====1
I

interaction and functional interdependence I

between these two. All kinds of technological
systems regardless of degree of complexity
may be part of a man-machine system, e.g.,
industrial plants, vehicles, manipulators,
prostheses, computers or management infor~a­
tion systems. For the interaction with such
systems, mostly psychological but also social
aspects are of concern. Task categories like
controlling and problem solving describe typi-
cal human activities in man-machine systems.
Later on, these task categories will be ex-
plained in more detail.
The overall purpose of any man-machine system
is to provide a certain function, product or
service as an output with reasonable costs,
even under conditions of disturbances influ-
encing man, machine or both (see Fig. 1). The
main goals or inputs of a man-machine system
are expected values of performance, costs,
reliability, and safety. At least since some
spectacular accidents have occurred with air-
craft and nuclear power plants, reliability
and safety have become vitally important op-o
op_o
erational as well as design goals in addition
to performance and costs; see Sheridan (1982).
Also, an acceptable level of workload and job
satisfaction of the man should be maintained;
Horay (1979).
see, e.g., Moray
I I
L ____ _ _ _ _ _ _ _ .JI
______________________________
Disturbances Disturbances
Fig. 1. Man-Machine
Han-Hachine System (HMS)
Some of these goals are in conflict with each
other. Such conflicts have to be resolved in
the most favorable manner by the designers
of a particular man-machine system. Any defi-
c~enc~es left to the human user of such sys-
C1enC1es
tems may cause poorer performance, job satis-
faction, and safety.
The interaction between man and machine is
the essential aspect of a man-machine system.
Classical ergonomic aspects like knobs and
dials design, anthropometry, lighting, or
adverse environmental factors have intensively
been investigated. Many
Hany results are available,
although not always applied appropriately. In
contrast, the focus of attention has centered
on inf~rmational aspects in the last years.
Questions of concern to a successful inter-
action between man and machine are:
What kind of information is needed?
How should the information be organized?

X111
x~~~
XlV
x~v G. lohannsen
Johannsen

Which information should be preprocessed? respect to human use, these were designed
intuitively by experience. This is even today
How should the information be transmitted?
a very common method. With more complex and
faster responding technological systems how-
All of these or similar questions can arise
ever, it turns out to be more and more man-
in different application areas. The questions
datory in many application areas to use ana-
relate to the control of technological sys-
lytical and consciously applied methodologies
tems, namely, to the degree of automation as
and systematic techniques for the design of
well as to the design of computer-generated
the man-machine system as a whole.
displays with preprocessing capabilities.
Further, they relate to all kinds of human-
For about 40 years, methodological knowledge
computer interaction as well as to manage-
has been gathered and systematic techniques
ment tasks on different organizational levels.
have been elaborated. Most of the first in-
The importance of the shifting from hardware
vestigations of man-machine systems were con-
and environmental aspects to software con-
ce~ned with manual control tasks, often ap-
siderations is nowadays expressed by the new
pl~ed
plied to aircraft piloting, later also to
term software ergonomics.
ship steering, car driving, and industrial
process control. This work was done either
From the preceding discussion, one can see
by experimental psychologists or by control,
that a wide range of technical areas is in-
systems, and application-oriented engineers.
volved and contributing to the field of man-
Overviews and literature surveys have been
machine systems. Therefore, a conference
given in several books: Kelley (1968), Oppelt
dealing with the subject will necessarily be
and Vossius (1970), Edwards and Lees (1974)
interdisciplinary in nature. Consequently,
Sheridan and Ferrell (1974~, and lohannsen
Johannsen '
this conference is sponsored by four inter-
etal.
et a1. (1977).
national federations which represent the
most important disciplines concerned with the
Many control theoretic models were developed
field of man-machine systems, namely IFAC
to describe the behavior of the human opera-
(automatic control), IFIP (information pro-
tor in manual control tasks. They have suc-
cessing), IFORS (operational research),
research) and
cessfully been applied as design tools for
lEA (ergonomics). '
automatic control systems which are better
adapted to the human operator, for unburden-
ing displays, etc. The most sophisticated
HISTORICAL AND SCIENTIFIC
and well validated model is the optimal con-
BACKGROUND
trol model shown in Fig. 2 in its basic form'
form;
see also, e.g., lohannsen
Johannsen and Govindaraj ,
A brief outline of the historical and scien-
(1980) and Pew and Baron (1982). It is struc-
tific background of man-machine systems may
tured into (1) a perception and attention
further illustrate the growing importance of
allocation part, (2) a central information
the field. The first existence of man-machine
processing part with an internal represen-
systems can be traced back into the early
tation of the system to be controlled, and
days of simple machines powered by men. With

white noise

r-
r -------
----
----
-----
----
--- ----------- - - - - - - - -----,
I I
Machine I
I
I
Disturbance· I
Filter I
I
I
I
control motion
I
control variables observable
observable variables
variables

l! It I I t It I
I
L_
L __
____
____
____
____
____
____
____
____
____
____
____
____
____
____
____
____
____
_J J

:------------------Hurnan-Operator-Model---- ----- -- --- ----,

1------------------------------------------
I Human Operator Model ---------------,
I
I Intended
intended

,.---...., \\r----
,....---
I control motion
I ,....--......
~~=t
I
I
I
II motor observation
I
I nOise
noise I noise
nOise I
~ ________ ~~~~
L ~c:.!~~ _ _ ___ __ _ _~~__Central
________ ~~~~~n~o~~~i~~
_____ Information
______ ~~C!~S~~
Processing
______ -L l'-=r.::?~~
_ ...l. _ _ _ _ _ _~
Perception _~t.!:~i~n_A~o.:.a.!i~_j
& Attention
_____ Allocation
_ _ _ _ _ JI

Fig. 2. Optimal Control Model

 Man-Machine Systems - Introduction and Background
(3) an action part generating optimal control
signals with respect to a cost criterion and
based on the estimates of all systems states.
With slower responding systems like ships and
industrial process plants, it became obvious
that it is more difficult to explain the hu-
man operator behavior by well established
control theoretic methods. The human control
behavior is highly nonlinear and intermittent
in these cases. Intrinsic monitoring, deci-
sion-making, and supervisory control behavior
became evident and attracted the attention
of several investigators; see Sheridan and
Johannsen (1976). Some methodologies from
operations research like network analysis
and queueing theory have been adopted and
comprehensive extensions of the optimal con-
trol model have been developed to describe
broader human operator tasks as well as the
whole design process for complex man-machine
systems; see, e.g., Siegel and Wolf (1969),
Pritsker and Pegden (1979), Rouse (1980),
Horaal and Kraiss (1981), and Pew and Baron
Moraal
(1982).
Another root for the field of man-machine
systems came out of what is nowadays called
cognitive science, a combination of cogni-
tive psychology and computer science. These
sciences developed without any strong rela-
tionship to man-machine systems. Hodels
Models of
the brain, theories for memory and thought
as well as human and artificial intelligence
and problem solving have been investigated;
see, e.g., Newell and Simon (1972), Klix
(1979) .
Task analyses show that problem solving tasks
are more important than control tasks in many
man-machine systems. Therefore, methodologies
from cognitive sciences have been adopted for
the analysis and design of these systems, but
only since a few years ago; see Rouse (1982).
Technological advances such as computers and
electronic displays have changed and will
continue to change man-machine systems in
almost all application areas. This is true
for industrial plants used for production or
power generation as well as for vehicles and
transportation systems. In addition, office
systems and information systems for observa-
tion, management, and command tasks in busi-
ness, defense, and medicine are today simi-
arly influenced. Not only the operation of
technological systems by highly skilled per-
sonnel, but also its use by inexperienced
people like in mass transport, as well as the
maintenance and design of systems are aided
by computers.
A common problem to all these applications is
the design of the human-computer interaction;
see,
see, e.g., Rouse (1981), Hatvany and Guedj
(1982), Williges and Williges (1982). The
possibly adaptive task allocation between man
and computer, the dialog design including
including the
use of
of natural language, and other software
software
ergonomic
ergonomic aspects are
are nowadays especially im-
portant topics of research and
and development.
xv
Also, the design of knowledge-based systems
will lead to helpful tools in such areas as
computer-aided decision-making, information
retrieval, and fault diagnosis.
With more computerization and higher degrees
of automation, new social effects and per-
spectives become evident. The advanced tech-
nologies allow a more flexible work organi-
zation with higher user acceptance and job
satisfaction. However, this advantage can
only be achieved if the social implications
are considered early enough by the designers
of future computerized man-machine systems;
see Margulies
Hargulies and Zemanek (1982).
HUMAN
HUHAN TASK CATEGORIES
All tasks of human personnel in man-machine
systems can be condensed into only two cate-
gories:
(1) controlling and
(2) problem solving.
These human task categories are fairly gen-
eral. Fig. 3 shows an attempt to integrate
them into a schematic block diagram.
Controlling shall here be understood In
~n a
broader sense than, e.g., in control theory;
see also Johannsen and Rouse (1979). It comp-
rises controlling in the narrower sense (in-
cluding open-loop vs. closed-loop and linear
vs. intermittent controlling) but also all
other action-oriented tasks such as reaching
and discrete-event acting (e.g., switching,
typing). Only through controlling, outputs
of the man-machine system to the environment
can be produced (see Fig. 3).
In contrast to controlling, problem solving
is an internal process on a higher cognitive
level. It comprises different tasks, mainly,
fault managing (especially fault diagnosing)
and planning. Fault managing is concerned
with solving problems in actual failure situ-
ations, thereby using and updating certain
rules with the objective of returning to a
good state of the overall system; see also
Rasmussen and Rouse (1981) and Johannsen
(1981). Planning is concerned with solving
possible future problems in the sense of men-
tally generating a sequence of appropriate
alternatives or rules for reaching future
states under different foreseeable and un-
foreseeable conditions; see also Johannsen
and Rouse (1982). In all these problem solv-
ing tasks, the rules are stored in the knowl-
edge base after their generation or modifi-
cation. From there, they can be utilized in
the lower-level process of controlling (see
3).
Fig. 3).
All other tasks in man-machine systems such
communicating, can
as, e.g., monitoring and communicating,
classified as subtasks or
be classified or supporting
controlling and problem
tasks of controlling problem solving.
two very different
Communicating comprises two
communication between the members
types of communication
of the group jointly responsible within aa
xvi G. Johannsen
J ohanns e n

r -- - - -
- -
- - - - - - - - - - - - - - - -- - - - - - - - - - - - - -
- -
- - - -.,
,
I
Model
I

Goals
Selection
Select ion 1 [ 11
,I
I
I
I
Generat .on&
Generatron& I
Interpreting Problem Knowledge I

~
I
of Inputs Solving Modlflcotlor
Mod rflcotlor Base I
I
I
I
I
Util,zation
Utili zation I
11 I

Selection 2 [K r---------------'
r
I
I
I,
- - - - - -- ------- J
I

Perception
Percep t ion ActIons
Ac t Ions Technological Consequenc es IOutpu
Consequences I Outpu 15
Controlling ,,
~
I
of Deviations I System
I
I
I

,
I
L_
_------------------ -- - -- - ---'
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ JI

DIs
Distturbances
urbances

Fig.
Fi g. 3 . Controlling
Contr ol ling and Problem
Prob l em Solving

man-machine
ma n-machine system.sys t em . The second type t ype iiss the t he Man, Cyberne
Cybernetics ti cs , SMC-l 0, 0 , 249 - 261
26 1..
communication
communi ca ti on of the t he man with tthe he ttechnolo-
echno l o- Johannsen,
Joha nn se n , G. (1981) ( 198 1) . Human- computer inter- i nte r-
gical
gi ca l system
s ys t em and the environment,
e nv ironme nt, sometimes
some times ac
actioti on n in decentralized contro controll and
identical
i de nti ca l with the human-computer
huma n-compu t er dialog.
di alog . It fault management of dynamic systems.
includes sensing,
se ns ing, perceiving,
pe r ce i v i ng , and interpret-
i n t e r pr et - Proc . IFAC 8th Triennia l World Congress ,
ing of o f input
inp u t information
info rma ti on and has been con- Session
Sess i on 74, Kyoto Kyoto,, August . Pergam Pergamon on Pr ess ,
Press
ssidered
ide r e d by Se Selection
l ec ti on 1 and Se Sell ec ti
tion
on 2 in Oxford.
Oxfo r d.
Fig.
Fig . 3. The Th e man
ma n se le l eccts
ts only
onl y those
t hose inputs
inpu t s Johannsen,
Johannsen , G., and W.B. . B. Rouse (1982).
( 1982) . Studies
from
fr om the ge generally
ne r a ll y available
ava il ab le information
i nforma ti on of planning
pl ann ing behavior of aircraft pilots
source
sour ce which
whi ch he wantswant s to t o use.
u se . This
Thi s selection
se l ec ti on in normal,
normal , abnorma
abnormall,, and eme r ge nc y situ-
emergency
is
i s handled
handl e d differently
diffe r e ntl y for f or the purpose
pur pose of ations . IEEE Tr Trans
ans.. Systems , Man, Cyber -
problem
prob lem solving
so l ving where
whe r e inputs are ar e interpret
i n t e r pre t ed , net ic i css (submitted).
(submitted) .
and theth e purpose
purpos e of cont controlling
ro lling where, at at a Kell~R.
Ke ll~R . ( 1968). 1968) . Manual
Ma nual and Automatic
much
mu ch higher
hi gher pace,
pace , deviations
dev i a ti on s are perceived.
pe r ceived . Control.
Con t rol . Wi l ey , New York. York .
Klix, F. (Ed .) . ) (1979).
( 1979) . Human and Artificial
Many of thet he topics
t opi cs mentioned in this thi s intro-
int ro- In te llige
I ntel nce. North Hol
li gence. Holland,
l and, Amsterdam.
duction will
du c ti on wi ll be ffur therr eelaborated
urthe l abor a t e d in tthe
he Margulies,
Margu lies, F., F. , and H. Zema nek (198
Zemanek ( 1982)2) . Man ' s
survey pape
papers
rs and throughout
thr oughout the seven ttopic opi c role
ro l e ini n man- machine systems.systems . Proc . IFAC IFAC//
areas
ar eas of this conference
confe r ence . IIFIP /IFORS/IEA Conf . Ana
FIP/IFORS/IEA l ysis , Des i gn,
Analysis
and Evaluation
Eva l uat i on of Man-Machine
Man - Mach i ne Systems ,
Baden- Baden, Sept . Perg Pergamon
amo n Press, Oxford .
REFERENCES Mo r aa l, J.,
Moraa J. , and K.-F.K.- F. Kraiss (Eds.) (Eds . ) (1981)
( 198 1) .
Manned Sys t ems Design
Systems De sign - Methods
Methods,, Equip -
Edwards,
Edwa rd s, E., and F.P. Lees (Eds (Ed s .) (1974).
( 1974) . The ment,
ment , and Applications . Plenum Press ,
Pl enum Press,
Human Operator
Ope r a t or ini n Process
Proce s s Control.
Co ntro l. TaylOr New York.
and Francis,
Franc is, London. Moray , N. (Ed.) (Ed . ) (1979).
( 1979) . Menta
Me n ta l Workload
Work l oad : Its
Hatvany,
Ha t vany , J., and R.A. R.A . Guedj
Gue dj (1982).
( 1982 ). Ma Man-ma-
n-ma- Th eory and Measureme
Theory Measurement nt . Plenum Press,
chine interac
inter ac tion
ti on in comput
computer- er- aaided
ided dede-- New York .
sign
si gn systems
sys t ems . Proc
Proc.. IFAC/IFIP
IFAC/ IFIP// IIFORS/IEA
FORS/ I EA Newe lll,l , A. , and H. A. Si mon ( 1972). Human
Conf.. Analysis,
Conf Ana l ys i s , Design,
De si gn, and Eva lu luati
a ti on Problem
Prob l em Solving
Solvi ng . Prentice
Prent i ce --Hall,
Hall ,
of Man-Machine
Man- Machine Systems,
Sy st ems , Baden-Baden,
Ba den- Ba den , Sept
Sep t.. Englewood-Cliffs,
Eng l ewood - Cliffs, New JJersey ersey .
Pergamon Press,
Pre ss, Oxford.
Oxf ord . Oppe lt,
l t, W ., and G. Vossius
W., Voss i us (Eds .) . ) (1970).
Johannsen,
J ohanns en, G., H.E. H.E . BoIler,
Bo Il er, E. Donges,
Donge s , and Der Mensch als Regler. Regler . VEB Verlag Techn Technikik ,
W. Stein (1977).
(1977) . Der De r Mensch
Me nsch im Regel-Rege l- Ber lin.
l in.
kreis
kr ei s - Lineare Modelle.Mode ll e . Oldenbourg
Oldenb ou r g Ver-
Ver- Pew,, R.W
Pew R. W.. , and S. S . Baron ( 1982) . Perspec Perspecttives
i ves
lag,
l ag , Miinchen.
Miinch en. on human performance model mode lling
l ing . Proc
Pr oc .
Johannsen,
Johanns en, G., and W.B. Rouse (1979).
Rous e (19 79) . M Mathe-
a th e- IFAC/
I FAC/ IFIP /IFORS/IEA Conf. Ana l ys~D
I FI P/IFORS/IEA ys~Dee ­
ma ti ca
c all concepts
concept s fo forr mode ling hu human
man be-
be - sign,
s i gn, and Evaluation
Eva l uation of Man- Machi Machine ne Sys -
havior
hav ior in complex
c ompl ex man-machine
man-mach ine systems.
sys t ems. tems,
t ems , Baden- Bade n,, Sept . Pergamon
Baden-Baden Pe r gamon Press,
Press ,
Human Factors,
Fac t ors, 21, 2 1, 733-747.
733 -7 47. Oxford .
nns e n , G., and ~ Gov
Johannsen,
Joha Govindaraj
ind a r a j ((1980).
1980) . Op-
Op - Pritsker,
Pri t sker , A.A A. A.. B. , and C. C.D.
D. Pegde
Pegden n ( 1979) .
timal
tima l con troll mode
contro modell predictions
predi c ti on s of sys-sys - Intr oductii on to
I n t roduct t o Simulation
Si mulation and SLAM .
tems pe performance
r fo rmance and attention
a tt e ntion aallocation
ll oca ti on Hile
Wi ley, y , New York .
and the
theirir experimental
experiment a l validation
va lid a t ion in in a Rasmussen,
Rasmussen , J., and W.B. . B. Rouse (Eds . ) ( 198 1)..
1981)
display
dis p lay desdesign
i gn study. IEEE IEEE Tr Trans,
ans. Sys Systems,
t ems , Hu
Humamann De t ection
ec t ion and Diagnosis of System Sys t em
 Nan-Nachine
Han-Hachine Systems
Systems Introduction
Introduction and
and Background
Background xvii
xvii

Failures.
Failur e s. Plenum
Plenum Press,
Press, NewNew York.
York. Sheridan,
Sheridan, T.B.,
T.B., and
and G.
G. Johannsen
Johannsen (Eds.)
(Eds.)
Rouse,
Rouse, W.B.
W. B. (1980).
(1980). Systems
Sys terns Engineering
Engineering Nod-
Hod- (1976).
(1976). Nonitoring
Honitoring Behavior
Behavior and
and Super-
Super-
els
els of of Human-Na chine Interaction.
Human-Hachine Interaction. North
North visory
visory Control.
Control. Plenum
Plenum Press,
Press, New
New York.
York.
Holland,
Holland, New New York.
York. Sheridan,
Sheridan, T.B.
T.B. (1982).
(1982). Neasuring,
Heasuring, modeling,
modeling,
Rouse,
Rouse, H.B.
W.B. (1981).
(1981). Human-computer
Human-computer interac-
interac- and
and augmenting
augmenting reliability
reliability of
of man-ma-
man-ma-
tion
tion inin the
the control
control ofof dynamic
dynamic systems.
systems. chine
chine systems.
systems. Proc.
Proc. IFAC/IFIP/IFORS/IEA
IFAC/IFIP/IFORS/IEA
ACN
ACH Computing
Computing Surveys,
Surveys, 13,13, 71-99.
71-99. Conf.
Conf. Analysis,
Analysis, Design,
Design, and
and Evaluation
Evaluation
Rouse,
Rouse, H.B.
W.B. (1982). Hodels Of
(1982). Nodels Of human
human problem
problem of
of Nan-Nachine
Man-Machine Systems,
Systems, Baden-Baden,
Baden-Baden,
solving:
solving: detection,
detection, diagnosis,
diagnosis, andand com-
com- Sept.
Sept. Pergamon
Pergamon Press,
Press, Oxford.
Oxford.
pensa ti on for
pensation for system
system failures.
failures. Proc.
Proc. Siegel,
Siegel, A.I.,
A.I., and
and J.J.
J.J. Wolf
Wolf (1969).
(1969). Man-Ma-
Man-Ma-
IFAC/IFIP/IFORS/IEA
IFAC/IFIP/IFORS/IEA Conf. Analy~ De-
Conf. AnalySIS; De- chine
chine Simulation
Simulation Models.
Models. Wiley,
Wiley, New
New York.
York.
sign,
sign, and and Evaluation
Evaluation of of Nan-Nachine
Han-Hachine Sys-
Sys- Williges,
Williges, R.C.,
R.C., and
and B.H.
B.H. Williges
Williges (1982).
(1982).
tems,
tems, Baden-Baden,
Baden-Bade n, Sept.
Sept. Pergamon
Pergamon Press,
Press, Human-computer
Human-computer dialogue
dialogue design
design consider-
consider-
Oxford.
Ox f ord. ations.
ations. Proc.
Proc. IFAC/IFIP/IFORS/IEA
IFAC/IFIP/IFORS/IEA Conf.
Conf.
Sheridan,
Sherid a n, T.B.,
T.B., and
and H.R.
W.R. Ferrell
Ferrell (1974).
(1974). Nan-
Han- Analysis,
Analysis, Design,
Design, and
and Evaluation
Evaluation of
of Man-
Man-
~lach
Ha c hin
i nee Systems:
Sys terns: Information,
Information, Control-,-
Contro 1-,- Machine Systems,
Machine Systems, Baden-Baden,
Baden-Baden, Sept.
Sept.
and
and Decision
Decision Nodels
Hodels of
of Human
Human Performance.
Performance. Pergamon
Pergamon Press,
Press, Oxford.
Oxford.
MIT
MI T Press,
Press, Cambridge,
Cambridge, Nass.
Mass.