Sunteți pe pagina 1din 7

Transpn Re~'-A. Vol 25A.

NO 5, pp 3t9-325, 1991
Printed m Great Britain

0191-2607191 $3 00 + .00
1991 Pergamon Press p|c

ARTIFICIAL INTELLIGENCE TECHNIQUES


FOR URBAN TRAFFIC CONTROL
MAURIZ10 BIELLI
Istituto di Analisi dei Sistemi ed Informatica,
Consigho Nazionale delle Ricerche, Viale Manzoni: 30-O0185 Roma, Italy
and
GIORGIO AMBROSINO, MARCO BOERO, and MARCO MASTRETTA
AUTOMA s.c.r.l., Via al Molo Vecchio, Calata Gadda, 16126 Genova, Italy
Abstract-This paper deals with the applications of artificial intelligence techniques to urban traffic
control problems, with the aim of improving the performances of current signal plan selection systems.
In particular, an architecture of an intelligent traffic control system is outlined with regards to the
different levels of data collection, data analysis and interpretation, decision, and control. The functionalities of hybrid modules introduced are discussed and the artificial intelligence methods used are mentioned. Finally the ongoing research in the field is presented.

INTRODUCTION

Recently, there has been a growing interest in the


application of artificial intelligence (AI) technology
to road traffic engineering problems. A number of
surveys on the subject provide a broad view of ongoing and potential applications of AI and knowledgebased systems (KBSs) in this field (see for example,
Hendrickson et al., 1985; Yeh et al., 1986; Bonsall
and Kirby, 1986; Kaplan, 1987; Johnson, 1988). Specialised conferences on the use of innovative technologies in the field of transportation engineering include various presentations on the application of
knowledge-based and expert systems (e.g., 2nd Int.
Conf. on Road Traffic Monitoring, 1989; 1st Int.
Conf. on Applications of Advanced Technologies in
Transportation Engineering, 1989). Several meetings
focusing on this application area have been recently
organized, both in Europe and North America (e.g.,
SERC Workshop on Expert Systems in Transportation, 1987; IEE Colloquium on Applications of Expert Systems in Road Transportation, 1988; TRB
Workshop on Expert Systems in Planning and Control, 1990). Some major European Economic Communities (EEC) initiatives, like DRIVE, Prometheus,
and Europolis programmes, give a prominent position to the use .of AI and knowledge-based techniques in the development of new road transport informatics systems.
This paper addresses the application of AI ideas
and methods to the specific context of urban traffic
control problems. More precisely it focuses on the
use of AI techniques in the framework of current
urban traffic control systems (UTCs) technology. At

present, several aspects and problems of the different


functions of the systems, from data collection and
analysis, to prediction and control, suggest the exploration of innovative approaches through which
conventional solutions could be enhanced. AI technology has developed a set of tools and methods
that appear to be ripe enough to be considered for
application to such problems.
The paper will present both some general ideas
on the integration of AI in the framework of current
traffic control techniques and an overview on current
research efforts in this direction.
OPEN PROBLEMS OF UTC SY~I'EMS

Urban traffic control systems are implemented in


a variety of ways. Generally speaking, however, in
the current architecture o f such systems three major
levels of operation cut across the different realizations:
Level 1 - , Traffic Data Collection
Level 2 --, Data Analysis/Interpretation
Level 3 -~ Decision and Control

Given the complexity of the traffic phenomena, and


the complexity of the control options provided to
traffic operators, the management o f the operations
implied by these three levels is, as largely recognized,
a knowledge-intensive task. An improvement o f the
quality of decision-making process is expected at all
the different levels of this task, from data collection
and interpretation to the decision of the best control
action to implement. Such decisions involve considtWork partially supported by European Economic erable expertise and imply, for a great part, appficaCommunities in the framework of the DRIVE Project tion of rules concerning regulation and accepted
VI055, developed in cooperation with Forschungszentrum
practice within constraints.
lnformatik an der Universitat Karlsruhe, Germany, Heusch
For each one of the above levels, it is possible
Boesefeidt Gmbh, Germany, Senter for Industriforskning,
Norway.
to summarize some problems and the corresponding
319

320

M. BIELLI et aL

basic functionalities which can be improved by an


AI approach. Synthetically, we include the following
problems, with particular attention to the traffic actuated signal plan selection policy (which is today
the most widespread approach to UTCs in Europe).

Level 1 (data collection}


Because of both economical and technical factors
(e.g., cost of sensors, problems concerning installation, maintenance, etc.) generally not all sections of
the road network are equipped with detectors. Thus,
only a partial view of the whole traffic on the network is generally available for analysis and decision.
Often, however, more data are required to understand the actual traffic situation and to determine
the adequacy of a signal plan with regard to that
situation.
The general objective, at this level, is to improve
the overall quality of traffic data collected by roadside sensors. The basic task is to reconstruct the missing traffic data starting from the set of available
data, for example, (a) reconstruction of traffic data
of nonequipped sections of the road network; (b)
plausibility checking of traffic measurements (e.g.,
in case of detectors fault) and reconstruction of erroneous data; and (c) reconstruction of other kinds of
aggregated traffic data, not directly acquired from
roadside sensors, for example, calculation of turning
flows over several intersections.
The problem of the integration of traffic data
coming from heterogeneous information sources can
be also considered at this level. Image processing
applications have shown that a significant improvement relative to the quality (and quantity) of traffic
data can be obtained (see, e.g., Dickinson and Asworth, 1983; Blosseville et al., 1989; Shimizu and
Shigehara, 1989). Yet this implies that, in order to
achieve a fully automatic integration and processing
of image-extracted traffic data in the UTC operation
flow, some data fusion (Luo and Kay, 1989) techniques must be developed and used.
In this sense, the overall aim of this level might
be better stated as completion (both in terms of quality and quantity) and integration of acquired traffic
data.
Level 2 (data analysis and interpretation)
This functional level is essentially concerned with
the identification of the current traffic state and its
development on the whole network area. Basically
this implies (a) the analysis of the current traffic situation on the network (as "completed" in the above
sense); (b) the understanding of future developments, at least in a qualitative way (e.g., whether the
flow or the density along a given direction is likely
to increase/decrease in the near future); and (c) the
diagnosis of the recognized traffic states.
Some problems appear to be the most important:
(a) interpreting traffic measurements and classifying
current traffic states; (b) recognizing tendencies towards possible future inconvenient changes of state;

and (c) analyzing congested and precongested states


in order to diagnose their causes.
The first two points are closely related to the selection of signal plans. As this selection takes place
not only on the basis of local aspects, it is important
to understand the development of the traffic on the
whole network. Also, a qualitative estimation of
near future states is important (e.g., in order to delay/anticipate the selection of a given signal plan).
The last point forms the background information
for solving congestion. Given the interactions of the
flows on the network, typically congestion grows up
in the form of "patterns" of congested zones. In order to recognize the reasons for congestion, an analysis of traffic relations between neighbouring measuring detectors on the network and other influential
factors on traffic demands (e.g., time of the day,
weather, etc.) is required.
Incident detection (i.e., the capability of the system to classify some congestion phenomena as due
to the occurrence of road accidents or incidents in
general) could be also regarded as a special task of
this level.

Level 3 (decision and controO


The general aim, at this level, is to improve the
quality of decision-making in traffic control actions
management, that is, to estimate the consequences
of possible actions and to adapt the decisions to the
given and to the evolving traffic situations. Basically,
the following problems have to be faced: (a) estimations of the effects of various signal plan implementation. Until now, the method currently used is the
signal plan (SP) selection. The SP is chosen among
those computed offline and recorded in a library;
every SP in the library, in fact, is the outcome of
an optimization program facing predefined traffic
patterns. So, it is able to cope with a unique set of
flows on the network and, possibly, with patterns of
flows not far from this set of values. As an obvious
consequence, the SP to be actuated, selected according to the recognized reference traffic situation,
could not ensure its complete adequacy to the actual
traffic situation; (b) signal plan modification, as a
means of tuning the selected SP to better match the
actual traffic state; (c) analysis of the consequences
of switching procedures between different signal
plans; and (d) choice of the "optimal" signal plan
selection rate.
AN ARCHITECTUREOUTLINEOF AN INTELLIGENT
TRAFFICCONTROLSYSTEM
Keeping in mind the above problems, in this section a possible approach to the use of AI in online
control of urban traffic flows is presented. As it is
better explained in the next section, the AI modules
must be indeed understood as hybrid systems, in the
sense that they integrate classical mathematical programming and operations research methods and algorithms with AI techniques.

AI techniques for urban traffic control


Figure 1 shows a general sketch of the reference
architecture. Basically the system consists of: (a) two
main functional modules, AI-module#1 and AImodule#2; (b) an urban traffic domain model, that
is, a (nonconventional) Traffic Process Model module, TP module; (c) a Model Tuner module; and (d)
flow of information among modules.
Roughly, the basic functionalities of levels 1 and
2 are embedded in the AI-module#1. Similarly, the
functionalities of level 3 correspond to AI-module#2.
Both modules have their own knowledge bases of
rules and behaviours. Although with different purposes, both use the Traffic Process Model as a principal source of data for their reasoning.

321

basing the selection of the control action~ on a set


of reference states defined on statistical analysis of
traffic behaviour. Nevertheless, in real practice, it
seems that experienced UTC operators make profitable use of their own cognitive models of the current
situation, for example, for detection and short-term
prediction of problems and for planning appropriate
reactions. Accordingly, the TP module contains a
descriptive, quafitative representation of the actual
traffic, including not only information on the current flows, but also general information needed for
traffic control.
A basic idea is to exploit, for online control purposes, at least part of the knowledge used by traffic
engineers in the planning phase when they design (or
redesign) the traffic control system. Normally this
knowledge is not available for the automated cycle
of operations of the UTC. Nevertheless, it often
seems to be implicit in many decision-making activities of the human operators when trying to interpret
the actual traffic situation to solve a problem.

TP module
The Traffic Process Model module is the backbone of the system and provides for the background
knowledge concerning the ongoing traffic situation.
Most of the existing UTC systems does not embody
any explicit model of the evolving traffic, merely

Road Traffic Situation

ctuetors

~gulatton
CRoadslde
sensors ~

tore

'

Truffle
/
measurement~

'

Hybnd~odule
Hybrid module

=1

Current Trefflc Stete8


Ernptrlcal and Expert Knowledge

Volumes
Densities
Speeds

Tendencies
RunnJnc,~Signal
l PI~
Trefflc Slgne! Plen8

Network

structure
Statistical

TP-HndBla
Fig. 1. The overall architecture.

Deta

322

M. BIELLIet al.

In this sense, the TP module is intended to embody snapshots of the traffic process as it progresses
in time, reflecting, instant by instant and as close as
possible, the actual changes on the real network.
This description is based both on static and dynamic
descriptive attributes of the process, such as: (a)
structural information about the network: for example, geometry of the network, hierarchical structure,
capacity of single links, turning possibilities in each
node, additional restrictions like bus lanes, etc.; (b)
macroscopic traffic flows parameters: that is, a~:tual
volume, density, speed on links, etc., for each network element, both in values and tendencies (decreasing/increasing/stationary), acquired by sensors
or inferred (by AI-module#1); and (c) empirical
knowledge of specbCicphenomena: for example, relevance of a particular intersection during the different
time periods of the day or related to traffic composition, etc.; its attitude to propagate problems to
neighbouring network elements; what areas should
be given the highest priorities in solving congestion;
what type of traffic to expect in a given area at a
given time of the day, and so on.
On the whole, this forms an enriched picture of
the ongoing traffic, which extends the partial picture
as formed by the set of data acquired by roadside
sensors.

A l-module#1
The TP module is fed by the AI-module#1, which
provides the data completion functionality mentioned above. AI-module#1 has access to the set of
measured data coming from sensors and to the current state of the Traffic Process. Using such information as a set of constraints on the unknown d a t a that is, using the known data and the empirical
knowledge of what is likely to happen, given the
actual situation-AI-module#1 tries to infer the
missing information in order to maintain a complete
picture of the evolving traffic.
Furthermore, it classifies the situation with respect to some criteria useful for selecting the appropriate signal plan. With the rules usually employed
in present practice, exceeding a single threshold
might be sufficient to identify a different situation.
As this might not be an adequate interpretation, AImodule#1 uses more complex rules which take into
account other factors, like the time of the day, the
day of the week, weather conditions, etc., to adapt
the interpretation of traffic flows to changing conditions.

Al.module#2
Using the results of data completion and analysis,
AI-module#2 draws the conclusions in order to determine the best decision as regards the actual traffic
control action. More specifically, it has to: (a) assess
the results of analysis and diagnosis performed by
AI-module#1; (b) judge on the adequacy of the current active signal plan. In the offline calculation process that leads to different SPs, to each SP corresponds an index that assesses the capability of the SP

itself to react to the traffic situation used as input


value. This Performance Index (PI) (such as, e.g.,
the total or average delay time on the network, or
the degree of saturation on the junctions) turns out
to be a useful means for the decision process where
we can calculate the same PI on the current traffic
situation. The comparison between this current PI
and the one associated to the running SP provides us
for an element, fairly concise for the first level of
decision: the adequacy assessment of the currently
actuated SP; and (c) propose a new signal plan SP,
calculated with reference to the current SP. It is, in
fact, either a SP available in the library, or a modification of the SPj that tunes it according to the situation and the needs arising from the diagnosis.
In doing so, AI-module#2 has to perform timedependent reasoning; that is, not only the current
interpretation has to be taken into account, but also
(near) past developments, recognized tendencies,
past decisions, and so on.
In the passage from the data acquired by the sensors to the information used by this AI-module#2, a
different temporal logic has to be taken into account.
In order to get a classification of the situation, the
AI-module#1 uses several sets of data measured by
the sensors, provided that the overall traffic situation
changes with a slower rate than the single link flow
does. In the same way, this control module can heap
up the different information fed via the TP-module
to perform a strategic control, that is, it has the
chance not to react just on the change of the situation.

Model tuner
The TP module maintains a qualitative picture
of the evolving traffic. The Model Tuner is a
knowledge-based module which performs metareasoning on the Traffic Process Model. It represents,
to some extent, a learning component that compares
the traffic situation (as described by acquired data)
with the picture of the evolving traffic as elaborated
by the Traffic Process Model module. In principle,
the discrepancies between the current traffic situation and the situation predicted by the TP are used
to update the rules which are embedded in it. For
example, one can imagine a metarule, in the Model
Tuner, which encode the following knowledge: if input data suggest that it is a typical rush-hour traffic
at link L3 and the TP states that it is not, then update
the rules, for it is rush-hour traffic at link L3.
Ai METHODS AND TECHNIQUES

Essentially AI means knowledge processing. It


implies the use of models of human reasoning and
problem solving, that is, conceptual frameworks for
organizing domain knowledge and strategies for
applying that knowledge to construct a solution of a
problem. Current AI technology provides for several
of such tools. Some of them seem to represent appropriate solutmns to favour the implementation o f the

AI techniques for urban traffic control


proposed architecture. In particular, such techniques, along with the logic and declarative approaches of AI, contribute to fulfill all the requirements that, so far, the quantitative methods of
operations research have not been able to meet.
We must take into account, however, that an important part is still played by the classic optimization
algorithms (e.g., for the calculation of signal plans)
and, more generally, mathematical programming algorithms (e.g., for the completion and the management of numerical data like those acquired by the
sensors). In fact, the provision of new paradigms,
which can deal with problems and knowledge of
qualitative kind with declarative representation
schemes, cannot replace the quantitative and analytical methods and the numerical computation algorithms, which, till today have been developed successfully.
In this way, the task of AI is to manage the mathematical models, to set up the solving algorithms,
to solve unstructured subproblems with the aid of
heuristics and to control the whole decision process.
As a consequence of the above outlined integration of classic algorithms and AI techniques, the
modules in the control loop have to be considered as
hybrid systems.

R ule-based programming
It is the best method, today, for realizing systems
which capture expert knowledge about some domain. Production systems are the most common
rule-based systems. A problem solving model based
on production systems consists of: (a) a knowledge
base, that is, a set of rules, each of which is formed
by a precondition and a set of actions (e.g., P --* Q
& R or ifP then Q and R) and captures some heuristic of the expert; (b) one or more data bases that
contain information about the domain and about the
problem-solving state (the working memory); (c) a
control strategy, that is, a way for controlling and
directing the reasoning steps performed by a rule
interpreter.
The rule interpreter executes a recognize-act cycle
that repeatedly looks for production rules whose
condition part matches some data in the working
memory, activates that rules (according to some selection mechanism which is part of the control strategy), and executes their action parts.

Heurtstic and model-based reasoning


Production systems are a quite straightforward
way for encoding knowledge and are normally employed when the relevant knowledge is expressible in
terms of situation-action pairs. In the general structure presented above, they seem to be well suited for
many tasks. For example, to express the heuristic
knowledge which surround the interpretation of a
given set of sensor readings, or the knowledge used
by the experts when judging about the effectiveness
of a given signal plan with respect to the interpreted
situation. Knowledge which can be embedded in and
TR(A) 25:5-G

323

used as a set of rules is sometimes referred to as

compded knowledge.
Different ways of representing and using knowledge are often required for different reasoning tasks,
for example, for reasoning on hypothetical consequences of actions or for making diagnoses. Modelbased reasoning is a deeper form of encoding and
using the knowledge of the domain. It implies the
use of models that describe the underlying system
structure and its components interactions. Such descriptions can be, for example, functional or causally
oriented abstraction of the system one is reasoning
about. Model-based reasoning provides for the basis
to analyse and explain the behaviour of the system
that is modelled, to capture its temporal aspects, and
SO on.
The TP module is to be encoded in this way. The
objects which make up the Traffic Process, their attributes, and their relationships must be represented.
This includes both static elements (the topology of
the network) and dynamic elements (the state of each
element in terms of descriptive variables, like volume
and density and their trends; causal relationships between descriptive variables, e.g. "the increase of traffic volume on link#n and link#m is the cause of a
precongested state on intersection#p," etc.). Qualitative modelling theories (Bobrow, 1985) are the best
tool available today for representing physical systems and perform qualitative reasoning on their behaviour.

Constraint programming
Constraint programming is a quite recent AI programming paradigm (McDermott and Charniak,
1985). It is based on techniques for representing constraints among variables and to use such constraints
for propagating values among variables (constraint
satisfaction). Problems which are tractable by constraint satisfaction techniques are characterized by:
(a) a set of elements to be interpreted, usually arranged as a network of elements, that is, a plausible
value is to be assigned to the elements; and (b) a set
of constraints among adjacent elements restricting
the value assignment with respect to the adjacent
elements.
Given some initial assignment of values to a subset of elements, a constraint satisfaction mechanism
tries to achieve an assignment of values to all elements that is consistent with the constraints. Depending on the kind of values, one can have symbolic
or numeric constraint propagation.
This method for representing and processing the
information can be used for the reconstruction of
data on nonequipped links and intersections. The set
of elements are the links and the intersections represented in the Traffic Process. The constraints represent the structural knowledge of the network, acquired traffic data, signal plans, etc. Using
constraint propagation, starting with the actual state
(a starting assignment), a consistent present traffic
situation might be inferred.

M. BIELLI el QL

324

Table 1. Some research projects on AI application to road traffic control


Project

Domain

SAGE: Expert system for sour- Urban traffic control


veillance and control of conArea control
gestion in urban areas
Urban traffic control
INTEL: Expert system for
Isolated intersection
traffic signal control in realtime
Urban traffic control
SCII: A KBS development in
Isolated intersection
real-time traffic control
VANESA: A KBS for urban
traffic management

Expert system for strategic response to freeway incidents


Expert Systems for emergency
management on urban freeways
TRANZ: Expert system for
traffic control in highways
work zones

Developers

References

INRETS, France

Field trials
(Paris)

Forast~ and Scemama


(1987)

Univ. of Taiwan

Prototype

Chang & Tang (1989)

Univ. Arizona, USA Prototype

Urban traffic control


Area control

Univ. Valencia,
ETRA (Valencia)
Spain
Incident detection and
Univ. Minnesota,
warning
USA
Safety and traffic moni- Umv. Roma, Italy
toring
Traffic control
Highways

AURA: Expert system for traf- Traffic control


fic control in urban motorways Motorways
TICC: Automatic incident reports interpretation and user
information broadcasting

Development
Stage

Prototype

Prototype

Lakshminarayanan &
Stephanedes (1989)

Feasibihty
study

Filippi et al. (1989)

Vtrgmia Dept. of
Prototype
Transportation,
USA
Univ. Madrid, Spain Prototype

Warning messages man- Univ. Sussex, UK


agement and broadcasting

ONGOING RESEARCH WORK

While the research efforts devoted to the introduction of AI concepts into conventional traffic control systems are comparatively few with respect to
similar efforts in other industrial application areas,
nevertheless, as already mentioned, in the last few
years some major research work has been undertaken along with this direction. Some of such work
is being carried out in the framework of coordinated
European research programmes. To give a rapid
overview of the current research on this topic, some
of the major projects are briefly described below.
DRIVE
The DRIVE progamme started in January 1989
and is currently funding about 50 research projects
aimed at the development and the application of new
road traffic informatics systems for improving traffic efficiency and safety. A wide range of aspects
are considered under DRIVE, from traffic control to
modelling and simulation, safety, impact evaluation,
and so on. Several projects dealing with specific issues (like, e.g., incident detection or user information and warning systems) make some reference to
the use of AI and KBS concepts. Three projects are
explicitly concerned with the analysis and the development of AI methods to be applied to traffic engineering and control problems.
One of these three projects is a feasibility study
of AI potentialities with respect to road traffic and
transportation engineering and has the final objec-

Radwan & Goul (1989)

Faghri & Demetsky


(1989)
Cuena (1989)

Demonstrator Allpon (1988)


Prototype

tive of compiling a selection of possible applications


of artificial intelligence in this field.
The two other projects aim at developing AI and
expert systems techniques to be used in urban traffic
control operations. Both projects address the development of AI-based extensions to current UTC technology and both are expected to deliver some prototype systems and on-field experimentation of the
concepts developed by the research.
Prometheus
In the framework of the Prometheus programme,
a specific subprogram (PRO-Art) is dedicated to the
research on finalized applications of artificial intelligence to the road traffic problems domain. Essentially P R O - A r t looks at the primary natural source
of traffic behaviour: the driver. The overall research
framework is specifically car-oriented. It aims at developing on-car AI systems, that is, AI-based functionalities to support the driver in managing the complex on-board facilities which are to be developed by
Prometheus (i.e., vehicle-to-infrastructure communication systems, vehicle-to-vehicle communication
systems, route guidance systems, etc.) and which are
expected to have a deep influence on traffic dynamics.
Basically, the P R O - A r t projects aim at developing some basic functionalities of an intelligent codriver provided with (a) planning and manoeuvre capability, that is, the vehicle's ability to regulate its
trajectory and speed, within the framework of an
overall plan, taking the positions and movements of

AI techniques for urban traffic control


the other vehicles into account and cooperating with
them as appropriate; and (b) a dialogue capability,
that is, the ability of the vehicle as an integral system
to maintain a satisfactory dialogue with the driver
through intelligent information presentation, natural
language and so on.
To achieve such functionalities, a number of research subareas are addressed by the different national research projects. Such areas appear to involve
a substantial body of AI base research interests and
methods: artificial vision, multisensor data fusion,
knowledge representation and planning, cognitive
modelling, man-machine interaction.
Beside coordinated research programmes and
studies, like DRIVE and Prometheus, there are several other projects dealing with the application of AI
to road traffic problems, involving both university
and industry teams. To provide just a sketchy view
of some of them, Table 1 summarizes those which
appear to be the most interesting from the perspective of traffic control.
CONCLUSIONS
We have presented some basic ideas for applying
artificial intelligence methods to urban traffic control problems. Some basic problems of current technology have been highlighted and some ongoing research have been briefly mentioned. No doubts the
research on the field is still at an infant stage, and
nevertheless is showing some promising developments. Artificial intelligence and knowledge-based
systems, in particular, are expected to give a significant contribution to the development of the next generation of urban traffic control systems (Bell, 1988;
Chase and Hensen, 1989).
Acknowledgements- The ideas presented in this paper are
based on the work currently being pursued in the context
of DRIVE project V1055. The authors would like to thank
all members of the project team: S. Fleischmann (FZI, Karlsruhe); R. Hock, H. Klrschfink (HB, Aachen); M. Richter
(HB, Hamburg); F. Melcarne (IASI, Roma); and D. Dewan, M. Irgens (SI, Oslo) for their valuable contribuuons.
A special thank you is also devoted to G. Improta and G.
Bruno (University of Naples, Dipartimento di Informatica
e SistemlstJca) for provided support and fruitful discuss~ons. The authors are also grateful the anonymous referee
for helpful comments.
REFERENCES

Allport D. (1988) Interpreting incident reports. IEE Colloquium on Apphcation of Expert Systems In Road
Transportation, Preprints. IEE, London.
Bell M. C. (1988) The fundamental issues of an expert system for urban traffic control, lEE Colloquium on Apphcatlon of Expert Systems in Road Transportation,
9(7), 1-6. lEE, London.
Blosseville J. M., Krafft C., Lenoir F., Motyka V., and
Beucher S. (1989) TITAN: A traffic measurement system using image processing techniques. IEE 2rid Int.
Conf. on Road Traffic Monitoring, Preprints. IEE,
London.
Bobrow D. G. (1985) Qualitative Reasoning about Physical
Systems. MIT Press, Cambridge, MA.

325

Bonsall P. W. and Kirby H. K. (1986) The role of expert


systems in transport. In P. Bonsall and M. Bell (Eds.),
Information Technology Applications m Transport, pp.
353-382. VNU Science Press, Utrecht, NL.
Chang A. T. and Tang R. Y. (1989) A prototype expert
system for traffic signal control in real-time, lit Int.
Conf. on Apphcations of Advanced Technologies in
Transportation Engineering, Preprints. San Diego, CA.
Chase M. J. and Hensen R. J. (1989) Traffic control systems- past, present and future. 1st Int. Conf. on Applications of Advanced Technologies In Transportation
Engineermg, Preprints. San Diego, CA.
Cuena J. (1989) AURA: A second generation expert system
for traffic control m urban motorways. 9th Int. Workshop on Expert Systems & their Apphcatlons, pp. 145156, EC2, Nantero, FR.
Dickinson K. W. and Asworth R. (1983) Traffic data collection and analysis using video processing. TEC,
Faghri A. and Demetsky M. J. (1989) TRANZ: A prototype expert system for traffic control in highway work
zones. 1st Int. Conf. on Apphcattons of Advanced
Technologies in Transportation Engineering, Preprints.
San Dxego, CA.
Filippi E., Guerriero G., Cecconi C., and Mantovani M.
(1989) Traffic monitoring and informauon technology:
The case of the Perugian urban freeway. IEE 2nd Int.
Conf. on Road Traffic Monitoring, Preprints. tEE,
London.
Forast~ B. and Scemama G. (1987) Intelligent traffic sourveillance system. Compendium of Technical Papers ITE
5th Annual Meeting, Preprints. ITE, New York.
Hendrickson C. T., Rehak D. R., and Fenves S. J. (1985)
Expert systems in transportation systems engineering.
Research Report, Dept. of Civil Engineering, CMU,
Pittsburgh, PA.
Johnson W. F. (Ed.)(1988)Proc. of the 2rid Workshop on
the Application of Artificml Intelligence to Transportation. Transport Canada, R&D Directorate, Tower "C",
Ottawa, Canada.
Kaplan E. W. (Ed.) (1987) Expert Systems for Transportation Apphcation, Transportation Research Board,
Washington.
Lakshminarayanan N. M. and Stephanedes Y. J. 0989)
Expert system for strategic response to freeway incidents. 1st Int. Conf. on Applications of Advanced
Technologies In Transportation Engineering, Preprints
San Diego, CA.
Luo R. C. and Kay M. G. (1985) Multisensor integration
and fusion in intelligent systems. IEEE Trans. Syst.
Man Cybern. 19, 901-931.
McDermott D. and Charniak E. (1985) Introduction to Artificial Intelligence. Addison Wesley.
Proc. of the 1st Int. Conf. on Apphcattons of Advanced
Technologies in Transportation Engineering. (1989) San
Diego, CA.
Proc. of the Colloquium on Apphcation of Expert Systems
in Road Transportation. (1988) lEE, London.
Proc. of the 2nd Int. Conf. on Road Traffic Momtormg.
(1989) lEE, London.
Provisional Report on SERC Workshop on Expert Systems
In Transport. (1987) ITS Univ. of Leeds, UK.
Proc. of the 69th TRB Annual Meeting Workshop on Expert Systems in Planning and Control. (1990). Transportanon Research Board, Washington.
Radwan A. E. and Goul M. (1989) Knowledge-based system development in real-time traffic control. 1st Int.
Conf. on Apphcations of Advanced Technologies in
Transportation Engineering, Preprints. San Diego, CA.
Shimlzu K. and Shigehara N. (1989) Image processing systems used cameras for vehicle surveillance. 2nd Int.
Conf. on Road Traffic Monitoring, Preprints. London.
Yeh C. l., Ritchie S. G. and Schneider J. B. (1986) Potential applications of knowledge-based expert systems in
transportation and planning engineering. TRR, 1076.

S-ar putea să vă placă și