Featuring developments in Federal highway policies, programs, and research and technology.
|This magazine is an archived publication and may contain dated technical, contact, and link information.|
|Federal Highway Administration > Publications > Public Roads > Vol. 64 · No. 3 > Design Evaluation and Model of Attention Demand (DEMAnD): A Tool for In-Vehicle Information System Designers|
Design Evaluation and Model of Attention Demand (DEMAnD): A Tool for In-Vehicle Information System Designers
by Christopher A. Monk, M. Joseph Moyer, Jonathan M. Hankey, Thomas A. Dingus, Richard J. Hanowski, and Walter W. Wierwille
The goal of in-vehicle information system (IVIS) technology is to increase the safety, mobility, efficiency, and convenience of the motoring public. Examples of these technologies include in-vehicle navigation/route-guidance systems, advanced traveler information systems (ATIS), and collision-warning systems. While the deployment of these technologies will help to reduce the number of crashes and fatalities on our highways, there is some concern about the potential for these systems to add to the problem of driver distraction.
Driving a vehicle imposes a particular load on drivers' attentional resources. Attentional resources can be thought of as a pool from which all tasks and mental activities are drawn. These attentional resources are used to safely perform the primary task of driving the vehicle (which includes vehicle control, navigation, and hazard detection). Interaction with an IVIS can increase the load on these attentional resources, possibly interfering with the driver's ability to perform the primary task of driving. Therefore, the design characteristics of an IVIS affects not only the amount of driver attention needed for the IVIS, but also the amount available for the driving task.
A critical aspect of designing systems for dynamic environments such as driving or flying is the amount of visual and cognitive attention required to complete a task using such a device. This required attention is termed "attention demand" by researchers and designers. For static environments such as desktop computing, attention demand is more important for usability issues than for safety issues. However, for devices in vehicles, the safety aspects of attention demand are paramount. Whereas failing to adequately address attention demand issues in a desktop software program may lead to poor usability, user confusion, and loss of revenue, failing to address these issues when designing an IVIS may result in loss of life or serious injury. Clearly, any guidance that can be offered to designers on how to reduce the attention demand of their devices would lead to safer systems and highways.
The Federal Highway Administration (FHWA) recently published Human Factors Design Guidelines for Advanced Traveler Information Systems (ATIS) and Commercial Vehicle Operations (CV0) (Publication No. FHWA-RD-98-057). These guidelines are based on extensive analytical and experimental research, and they are currently being used by ATIS and IVIS designers to address factors that can lead to complicated and demanding interfaces.
While design guidelines are extremely useful to designers, they do not allow designers to evaluate prototype designs for usability and, more importantly, attention demand. Researchers have long used mathematical models based on given parameters to predict various outcomes of systems. For example, traffic engineers can build models that predict throughput for an intersection design at a given location. In the case of predicting the attention demand of interfaces, the public domain has no formal models that are specifically tailored for in-vehicle devices.
In 1996, FHWA initiated a research project with two main goals: (1) provide designers of IVIS technologies with a set of tools and criteria that could be used in evaluating the attentional resources required by IVIS designs and (2) provide highway planners and engineers with tools and criteria to evaluate proposed IVIS requirements. More specifically, the desired outcomes of the project included a behavioral model that predicts the performance of drivers interacting with an IVIS and a prototype software package that uses the behavioral model to evaluate the attention demand required to operate a given IVIS. The behavioral prototype software was called "IVIS DEMAnD" (In-Vehicle Information System Design Evaluation and Model of Attention Demand). This program can be installed and run in a Windows operating environment.
How Was the IVIS DEMAnD Program Developed?
The data used to develop the IVIS DEMAnD program came from three general sources: (1) an extensive literature review (including many FHWA-sponsored research projects), (2) contact with known practitioners, and (3) a set of four on-road field studies conducted by the research team specifically for this project. The purpose of the literature review and the discussion with human factors practitioners was to gather existing data on driver-task measures. A review of the existing/available data revealed several deficiencies. Field studies conducted using a given simulated IVIS in actual vehicle tests were used to supplement the existing data. The model equations and analytical tools used in the IVIS DEMAnD program were then developed from this real-world data.
It is important to note that an expert group of actual designers of IVIS devices was assembled to consult on the development of this model and software prototype. These experts came from automotive manufacturers and suppliers. By including some end-users of the product in the development of the model and prototype, FHWA ensured that the user's needs would be met and that the product would be customer-driven.
What Does the IVIS DEMAnD Program Do?
The purpose of the IVIS DEMAnD program is to assist designers and engineers to evaluate the demands placed on the driver's attentional resources by given IVIS designs and their associated tasks. More specifically, the program can be used to: (1) compare two or more candidate IVIS designs for performing the same task; (2) evaluate an upgrade for a current design; or (3) evaluate a given design, task, or subtask against a set of benchmark criteria.
The benchmark criteria are safety-related measures that indicate how driver performance will be affected relative to baseline driving with no in-vehicle tasks. A sample list of these measures and their critical values is shown in table 1.
The software prototype was designed to assist the user in developing a conceptual model of the driver as a collection of resources with a limited capacity. These resources include visual, auditory, supplemental information processing (i.e., complex cognitive processes beyond information extraction), manual, and speech. It was also important for the user to perceive secondary tasks, such as operating an IVIS device, as being potentially in competition with the primary task of driving the vehicle for these resources. Finally, it was important that the user understand that the amount of additional load placed on the driver by these tasks depends on numerous factors, including:
The prototype was designed so that the user could describe various in-vehicle information systems in terms of the tasks a driver might routinely perform. The prototype was also designed to allow comparison of the effects on driver demand of different tasks and different systems. The "effects on driver demand" refers to measures of interest to human factors design engineers that can be used to evaluate a given IVIS.
How Does the IVIS DEMAnD Program Work?
The evaluation of an IVIS begins with the user specifying the IVIS task(s). Figure 1 shows one of the initial program screens that helps the user to specify the driver resources that are involved in performing the task. As described previously, the task can draw upon one or more of the following five resources: (1) visual demand, (2) auditory demand, (3) supplemental information processing (SIP) demand, (4) manual demand, and (5) speech demand.
After the designer has selected one or more resources that define the task of interest, the designer then selects the task that most closely matches the task of interest from one of two task libraries. One library is based on tasks taken from the technical literature (figure 2), and the other can be a library based on tasks previously used by the designer.
If the task cannot be found in either task library (i.e., the task is an uncommon task and/or a task without data), the user can specify the characteristics of the task by comparing it to other more common tasks. An evaluation tool ("Wizard") guides the user through the process of specifying the various characteristics of a task that is not in the libraries. This process of specifying task characteristics, such as the required mean single-glance time, is shown in figure 3. The user can specify a single value or an upper and lower bound on a value. The user can also specify a number of task characteristics (i.e., measures). A list of these characteristics is shown in table 2.
At this point in the evaluation, the user has specified: (1) the driver resource categories germane to the task of interest, (2) the task of interest or the characteristics of the task of interest, and (3) modifiers relevant to the design. Once these items have been specified, the user can view the results of the evaluation. As shown in figure 4, the evaluation is displayed graphically, and it illustrates the relative driving task performance (conceptual) and the degree to which driver resources are affected by the task. This conceptual driving task performance measure is called the figure of demand; it is a single overall measure that assesses the attention demanded of the driver. A Demand Measures Summary that outlines what measures are affected and the degree to which they are affected is also provided.
Through this model, designers not only have a prediction or assessment of attention demand for their interface designs, but they can also use the model as a diagnostic tool for determining which tasks or subtasks are significant contributors to high-attention demand. This ability to predict potentially demanding tasks and subtasks makes this software an invaluable tool for IVIS and ATIS designers.
What Is the Status of the Prototype Software?
The potential for the DEMAnD prototype software to have a substantial impact on the design of safer in-vehicle information systems is considerable. However, the current model of attention demand needs further development and, ultimately, validation before it can be used with confidence by designers. In addition, the task database needs more development and expansion before the DEMAnD software will be ready for wider distribution.
Currently, the prototype software has had a limited release to designers working for major automotive manufacturers and suppliers and to key researchers in the area of IVIS distraction and driver performance. FHWA has released the prototype software to these users with the expectation that they will respond with feedback on both the software (usability, functionality) and the model (validity, robustness, theoretical foundations). Once feedback has been assembled and analyzed, FHWA and representatives of the U.S. Department of Transportation's Intelligent Vehicle Initiative program will be able to determine the next course of action for the DEMAnD prototype software.
For more information about the DEMAnD prototype software, contact Joe Moyer at (202) 493-3370 or email@example.com.
Christopher A. Monk is a research psychologist with Science Applications International Corp. He works as an onsite contractor supporting the Human-Centered Systems Team's Intelligent Transportation Systems Program at the Federal Highway Administration's Turner-Fairbank Highway Research Center in McLean, Va. He has a master's degree in human factors psychology from California State University at Northridge and is currently pursuing a doctorate in human factors and applied cognition at George Mason University in Fairfax, Va.
M. Joseph Moyer is an engineering research psychologist and a member of the Human-Centered Systems Team in FHWA's Office of Safety Research and Development at the Turner-Fairbank Highway Research Center. He holds a master's degree in psychology from George Mason University.
Jonathan M. Hankey is the leader of the Advanced Product Evaluation Group at the Virginia Tech Transportation Institute in Blacksburg, Va. He holds a doctorate in industrial and systems engineering from the University of Iowa.
Thomas A. Dingus is a professor in the Department of Industrial and Systems Engineering and director of the Virginia Tech Transportation Institute in Blacksburg, Va. Prior to his appointment at Virginia Tech, Dr. Dingus was an associate director at the University of Iowa driving simulation facilities and was director of the National Center for Advanced Transportation Technology at the University of Idaho.
Richard J. Hanowski is a senior research associate at the Virginia Tech Transportation Institute in Blacksburg, Va. He earned a doctorate in industrial and systems engineering from the Virginia Polytechnic Institute and State University.
Walter W. Wierwille is senior transportation fellow and the leader of the Safety and Human Factors Group at the Virginia Tech Transportation Institute in Blacksburg, Va. Dr. Wierwille is a registered professional engineer in Virginia, and he is Paul T. Norton Professor Emeritus at Virginia Tech. He has a bachelor's degree in electrical engineering from the University of Illinois and a doctorate from Cornell University.
Page Owner: Office of Corporate Research, Technology, and Innovation Management
Scheduled Update: Archive - No Update
Technical Issues: TFHRC.WebMaster@dot.gov