Printer Friendly

Highway applications of expert systems.


Expert systems have gained acceptance in many fields, ranging from aerospace engineering to finance, but they have not gained widespread application in highway engineering. There are many areas where expert systems could be of substantial value in highways. This article will attempt to briefly summarize what expert systems are, current applications in highway engineering and operations, make observations about expert systems and their applications, and comment on current research and future use of expert systems.

Expert systems are computer programs that incorporate: (1) reasoning and problem-solving processes of human experts (heuristic problem solving) and (2) specialized knowledge of experts (experience). The goals of expert systems are usually more ambitious than those of conventional or algorithmic programs because expert systems frequently perform not only as problem solvers but also as intelligent assistants and training aids. Expert systems have great potential for capturing the knowledge and experience of current senior professionals (many of whom are approaching retirement age) and making their wisdom available to others in the form of training aids or technical support tools. Applications include design, operations, inspection, maintenance, training, and many others.

Existing operational expert systems have already demonstrated the feasibility of highway applications. These systems include FRED (Freeway Realtime Expert System Demonstration), which is a real-time prototype for managing nonrecurring congestion or urban freeways in Southern California, and ERASMUS, a pavement assessment and rehabilitation system that is operational on 35 sites in France. Other developed systems such as FASTBRID (Fatigue Assessment of STeel BRIDges) and WZTS (Work Zone Safety Training System) show that fully integrated decision-aid/training systems are both possible and practical.

Description of the Technology

Expert systems differ from conventional programs in the way they store and use information. In a conventional program, the operations never vary as they are predetermined by the programmer. The conventional program contains precisely defined logical formulas and data, and if any data element is missing, the program will not run. The expert system, like the human expert, contains heuristic information and can function with incomplete information. Some of the major differences between conventional programs and expert systems are show in table 1.


Current Applications in Highways

While applications in highway design, engineering, and operations have not gained the wide acceptance that expert systems have achieved in other fields, they are increasing in use. The results of a recent survey, conducted by the Organisation of Economic Co-operation and Development (OECD) in Paris, provided information on 90 expert systems in various stages of operation and development. This survey represents only a portion of expert systems actually developed.

The systems included in the OECD survey are classified by function in four very broad groups:

* Traffic management and control - systems developed to advise or assist with traffic management and control operations, such as diagnostics of traffic problems from sensor data, incident detection, signs, and signals.

* Traffic impact and safety - systems for evaluating ways of reducing the impact of traffic, such as noise control, safety work zone layout, accident investigation, etc.

* Highway design and planning - systems designed to assist with roadway design and to analyze roadway needs and problems, such as geometrics, landslide forecasting, and drainage.

* Highway management - systems to assist with roadway maintenance, operation, and decision-making, including pavement maintenance, bridge deck repair, and bridge painting strategies.

The systems reported were also grouped according to the category of problem. For the purposes of this report the categories were: (2)

* Diagnosis/monitoring. The basic goal of diagnosis/monitoring is to catalog a system's characteristics (deterioration, malfunction, etc.) into a specific cause or set of causes and from this develop solutions, i.e., what is wrong and what should be done about it. Monitoring can be considered to be real-time diagnosis. They are similar in terms of the problems and the complexities involved in developing the expert system.

* Interpretation/classifying. This class of system compares a situation with a set of known conditions and looks for matches. Expert systems that solve problems in this area are designed to model the pattern-matching ability of someone who is an expert in identifying features or characteristics in the problem domain.

* Prediction/forecasting. The goal of this class of system is to forecast future conditions based on existing conditions and a knowledge of past conditions.

* Design/planning. This type of system specifies how something should be built (design) or a prescribed set of actions to meet a goal (planning). In most examples developed to date, this consists of providing detailed specifications for a generic design or plan.

Table 2 shows the systems grouped by category and function. This table depicts expert systems that are in active use, in test and evaluation, or in development in 12 different OECD countries. All of the systems represented are either successful operational systems or potentially useful systems under development.


The heavy emphasis on highway management with diagnosis/monitoring systems and on highway design and planning with design/planning systems is apparent. This does accurately reflect needs in the highway community where funding and staff are often inadequate and the problems cannot be ignored or deferred.

Several observations and conclusions can be drawn from the responses to the OECD questionnaire.

* The expert systems reported appear to be more developer-driven than user-demanded. This is to be expected in any relatively young technology; however, there is gradual acceptance of expert systems by the user work force.

* PC-based expert systems are far more common than workstation-or mainframe-based systems. There are very few Macintosh-based systems. The reason for this situation is obviously the availability of PC's and PC-based development tools. There are, however, increasing numbers of expert systems being developed on higher level workstations and then being ported to PC's. Of the systems reported, about 80 percent were PC-based with the balance divided between workstations and mainframes (15 and 5 percent, respectively).

* The integration of knowledge-based expert systems with algorithmic systems and data bases and other technologies is firmly entrenched as a practice. Many of the existing systems are hybrid systems where the knowledge-based expert system interacts with external programs and data bases or is only one component of a larger system.

From the OECD survey (4), major expert systems conferences - in Paris (5), Espoo, Finland (6), Avignon, France (7), and Montreal (8) - and other sources, several additional observations can be made:

* Currently available development tools are adequate for building expert systems in both basic and complex technical areas of highway engineering and operations.

* Fully integrated decision-aid/training systems are both possible and practical by combining expert systems with interactive videodisc training systems and other conventional media.

* The time and cost of developing and implementing expert systems in high compared to the time and cost required for developing and implementing algorithmic systems of a comparable magnitude. The time and cost are expected to decrease as development tools mature and as procedures for the verification, validation, and evaluation of expert systems are refined.

Observations more specific to the problems faced during the development, testing, and application of expert systems include:

* Structured planning is recommended for the successful development of a system. This should include the problem/need and the system's benefits, organizational risk factors, technical risk factors, and user risk factors.

* Management support in the institution sponsoring the development of the expert system is necessary. This support must include the commitment of both staff and financial resources to succesfully develop and implement the system. Full knowledge and understanding of the costs, benefits, and risks is essential.

* The end user is pivotal to the development of expert systems and must be involved from the planning through the field evaluation stages. The end user provides definition of the skill level of the user community, information on how problems are addressed in practice versus the prescribed solutions, advice on how the system must function (interact with the user) to be accepted by the user community, and a cadre of supporters to test and promote the system once it is completed.

* Knowledge from the experts is vital throughout the development of the expert system. It is vital both in terms of building the system and for maintaining interest and continuity throughout the project.

* Reliability and ease of maintenace must be considered in all phases of the system development. Since the maintenance will probably not be performed by the developers, the system structure must be clear and straightforward. Logical and understandable names should be used for objects and knowledge structures within the system. Clear and complete doctem documentation is required for effective maintenance.

* The selection of the development tool for an expert system project should be performed by a qualified knowledge engineer or expert systems developer. This is critical because the tools have significant differences that are not explained in available literature and because an application should be keyed to the specific knowledge-handling and operational charateristics of a development tool.

Future Directions

Several of the expert systems currently under development should further demonstrate the value of expert systems and the variety of problems that can be addressed using them. For example, a small expert system is under development to diagnose signals from an inductive loop detector tester; when completed, the system will demonstrate the practicality of imbedding an expert system in testing hardware. Other current applications range from bridge rail retrofit design systems to pavement management systems to freeway incident management systems. There are numerous potential applications in the Intelligent Vehicle-Highway Systems (IVHS) Program.

One of the technical factors slowing the development and fielding of expert systems is the difficulty in testing these systems. There is little agreement among experts on how to accomplish verification (Is the system built right?), validation (Is it the right system?), and evaluation (Is the system valuable?) of expert systems. C. Green and M. Keys describe the vicious circle where "nobody requires expert system validation and verification, so nobody does it. Since nobody knows how to do it, nobody requires it." (9) One of the causes for this lack of agreement, and thus lack of accepted methodology, is the "combinatorial explosion" of possible solutions resulting from the execution of an expert system. The six-step solution proposed by J. Geissman and R. Schultz offers an approach to the validation and verification of expert systems, but it does not really address the complexity of the solution state space and offer processes to design field tests. (10) Fundamental research on these issues has been conducted through NASA, and applied research is being initiated by FHWA. (11)

New series of tools, including expert systems as one component, are in the planning phases. One example of such a hybird system is a voice-actuated tool to assist the construction inspector. The system will use a highly portable (wearable) PC as the host and incorporate a voice data input module, a visual display, a voice-generation module, and a report generator. The system is also expected to provide a pre-inspection refresher course to the inspector based on information contained in the expert system and stored graphic images.

Two areas of opportunity that are not receiving adequate attention are expert systems as training aids and intelligent data bases. The increasing potential for such systems, especially intelligent data bases, is generating a great deal of interest.

For a variety of reasons, not all of which are technical, expert systems have not achieved their potential in highway engineering and operations. However, the outlook for expert systems is quite optimistic, and the possibilities for greatly increased productivity are astounding.


[1] C.W. Schwartz. Lecture on expert systems at the Workshop on Expert Systems for Maintenance and Rehabilitation of Flexible Pavements, McLean, Virginia, 1988. [2] James A. Wentworth. "A Guide for Developing Knowledge-based Expert Systems," OECD Workshop on knowledge-based Expert Systems in Transportaion, Vol. 2, Espoo, Finland, 1990. [3] S. Linnainmaa. "Overview of Expert Systems Technology," OECD Workshop on Knowledge-based Expert Systems in Transportation, Vol. 1, Espoo, Finland, 1990. [4] Draft Summary Record of the Third Meeting of Scientific Expert Group RP2 on "Knowledge-based Expert Systems," DSTI/RTR/RP2/M(91)1, Organisation of Economic Co-operation and Development, Paris, France, 1991. [5] Colloque International "Route et Information," sponsored by L'Ecole National des Ponts et Chausses, Paris, France, 1990. [6] OECD Workshop on Knowledge-based Expert Systems in Transportation, hosted by the Technical Research Centre of Finland (VTT), Espoo, Finland, 1990. [7] AVIGNON91, Eleventh International Conference on Expert Systems and their Applications, Avignon, France, 1991. [8] OECD WOrkshop on Knowledge-based Expert Systems in Transportation: Operational Experience and Perspectives, hosted by Transport Canada, Montreal, Canada, 1992. [9] C. Green and M. Keys. "Verification and Validation of Expert Systems," workshop on knowledge-based system verification, NASA/Ames Research Center, Mountain View, California, April 1987. [10] J. Geissman and R. Schultz. "Verification and Validation of Expert Systems," Al Expert, pp. 26-33, February 1988. [11] M. Mehrotra and C. Wild." Multi-View Point Clustering Analysis," to be published in the 1993 Goddard Conference on Space Applications of Artificial Intelligence, 1993.

James A. Wentworth is the Acting Chief of the Operations Research Division, Office of Advanced Research. Mr. Wentworth joined FHWA in 1973 and has served in various technical and administrative functions in Research and Development. He recently returned to Turner-Fairbank Highway Research Center after serving as an Administrator for 2 years at the Organisation of Economic Co-operation and Development's Road Transport Research Program in Paris, France. From 1961 to 1972, he held various positions in private industry.
COPYRIGHT 1993 Superintendent of Documents
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Wentworth, James A.
Publication:Public Roads
Date:Mar 1, 1993
Previous Article:Pan American Partners: the Pan American Institute of Highways.
Next Article:Benefit-cost analysis of lane marking.

Related Articles
Technology transfers and its tools.
Technology transfer: American style.
Pan American Partners: the Pan American Institute of Highways.
TRB Superpave Committee: Keeping [Superpave.sup.TM] Implementation on the Road.
Strategic Work-Zone Analysis Tools.
New Standard Requires Internal Trunk Releases on Cars.
Internet Watch.
Making two-lane roads safer: IHSDM is an invaluable safety evaluation software package for highway designers and planners.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters