U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
|This report is an archived publication and may contain dated technical, contact, and link information|
Publication Number: FHWA-RD-95-153
Date: November 1996
Development of Human Factors Guidelines for Advanced Traveler Information Systems and Commercial Vehicle Operations: Literature Review
CONCLUSIONS AND RECOMMENDATIONS
The previous sections discussed products of primary interest for the initial literature review of this project. These products summarize existing guidelines applicable to ATIS/CVO systems, and a preliminary list of human factors research needs for guideline development.
PRELIMINARY GUIDELINE SUMMARY
Attempts were made to procure applicable human factors guidelines from several sources, including:
Each of these sources appears as a separate section.
In attempting to compile guidelines from these sources, it became apparent that there are literally thousands that apply to ATIS/CVO systems (particularly from existing guidelines). Therefore, the guidelines were prioritized and only the most applicable are discussed. As part of the final guidelines developed for this project, it will be necessary to pay close attention to a number of sources. The guidelines summarized below are provided in a list format by category of applicability.
Guidelines From Published ATIS/CVO Research
Driver attention and workload
Designers should allocate as many tasks as possible to "pre–drive" as opposed to "in transit" in order to minimize required driver attention. "Pre–drive" consists of the complex planning and attention–demanding tasks. "In transit" consists of a relatively small subset of tasks that are necessary for efficient system usage while the vehicle is in motion (Lunenfeld, 1990). Designers should limit in–transit functions to necessary tasks and those of major convenience. Tasks can successfully be completed without substantial driving task interference by properly selecting and designing in–transit functions (Dingus and Hulse, 1993).
Effort must be made to limit the functionality of the in–transit mode to those tasks that:
All other functions should be allocated to the pre–drive case or carefully considered (and tested) based on the above criteria (Dingus and Hulse, 1993).
Allocating functions to a "zero speed" category increases in–transit functionality without compromising driving safety (Carpenter, Fleishman, Dingus, Szczublewski, Krage, and Means, 1991). Zero speed means a vehicle is stopped and in drive. Once the vehicle starts to move again, the display and control configurations return to the previous in–transit state.
Selection of ITS interfaces nomenclature and labeling is critical to system usability. Nomenclature and labeling should be subjected to usability testing prior to final design of the system (Dingus, et al., 1991).
Pre–drive functions involve human–computer interaction for the computer illiterate. Three critical aspects generally apply to the automobile:
It is important to assess visual attention in driving since most information is gathered visually by the driver (Rockwell, 1972). Visual attention can change quickly, especially in certain circumstances (e.g., the presence of a curve, traffic, or a change in type of roadway), and can consume 100 percent of a driver's resources. To reduce attention demands on the driver, short display viewing times are necessary for driving at faster speeds and on more complex roads. Research has shown that gender and previous driving experience affect attention demands. Age also affects demands; older drivers spend significantly more time looking at displays than younger drivers (Pauzie, Marin–Lamellet, and Trauchessec, 1991; Dingus, Antin, Hulse, and Wierwille, 1989). Consequently, it is important to design information displays so that only the minimum glance time is required.
The display should be limited to only the necessary information, which, according to Streeter (1985), would only include the next turn, how far away the turn is, which street to turn on, and which direction to turn on a pre–specified route. Streeter found that people who are familiar with an area prefer to be given the cross street of the next turn, whereas people who are unfamiliar with an area prefer to be given distance information. In addition to proximal (i.e., next turn) route–following information, notice of upcoming obstacles or traffic congestion would also be beneficial. Such information could make the task of driving safer, given that it can be displayed without requiring substantial driver resources (Dingus and Hulse, 1993).
Traffic information, such as minimum travel time, route selection, and guidance, should be updated at least every 15 min. Information should be available to users in as many pre–drive options as possible to reduce in–transit workload. Options include telephone (including cellular phone), personal computer, videotext terminals, television, fax, and eventually on–board computer systems.
Voice presentation (particularly in situations of high visual–attention demand) can make driving easier and safer if designed properly (Dingus and Hulse, 1993). Dingus and Hulse (1993) recommended that:
As the quality of auditory messages decreases, the workload to process the messages increases. Although the quality of low–cost synthesized speech is constantly improving, factors such as tonal quality and inflection limit its effectiveness relative to digitized speech (Sanders and McCormick, 1987).
Even though research studies have been conducted to test forms of synthesized speech, the state–of–the–art knowledge is not yet to the point where intelligibility/comprehensibility can be predicted in all situations or environments. However, it is known that a number of factors influence intelligibility, such as speech rate, message length, message content, message complexity, background noise, pitch, and loudness (Van Cott and Kincade, 1972; Marics and Williges, 1988). Therefore, voice loudness, frequency, and spectral content must be carefully considered. Given the noise variation for in–vehicle environments, a voice volume control is essential (Dingus and Hulse, 1993).
In addressing some of the intelligibility concerns discussed above, Labiale (1990) recommended that designers restrict the amount of information presented (i.e., seven to nine bits) in aural messages, or use the aural cue as a prompt to a simple visual guidance presentation. Labiale also recommended that aural messages be repeated to aid in intelligibility and recall, especially if the information is complex.
Messages must be worded carefully so that drivers do not misinterpret them. A study conducted by a Japanese automobile manufacturer indicated that drivers tended to instinctively respond more to verbal information than visual information. Drivers tended to follow the in–vehicle instructions even if they conflicted with traffic regulatory information (e.g., turning the wrong way onto a one–way street) (Noy, 1991).
Route algorithms should display navigation information whenever possible. Without a planned route, drivers must plan trips in transit instead of pre–drive (Antin et al., 1990). Almost invariably, information displays for navigation information systems are (and probably will continue to be) quite small. Therefore, it is important to require that non–route systems provide "zoom–in," "zoom–out," and pan features in conjunction with prioritized streets to avoid unreasonably high screen information densities. This strategy, although necessary to make a display legible, increases driver attention, since pre–drive planning must (due to the difficulty of panning and zooming to plan many routes) be accomplished as an in–transit task (Antin et al., 1990). Dingus et al. (1989) therefore recommended that a provision for route selection be provided as part of navigation and information systems.
Another advantage to providing a route selection algorithm as part of the navigation system feature is that many more options are available for information presentation. If no route is provided, an area map must be displayed to navigate accurately. If a route is provided, the navigation information can be displayed aurally and/or visually, textually or spatially, as well as in a turn–by–turn graphic format or an entire route graphic format.
The issue of whether to use maps, turn–by–turn graphics, or textual direction lists is somewhat task–dependent when presenting visual navigation information. Studies (e.g., Streeter, Vitello, and Wonsiesicz, 1985) indicate that textual lists are easier to use than maps when navigating to unknown destinations. Note, however, that maps provide additional information (e.g., orientation information such as cross streets) that textual lists do not. Therefore, whether a map or list is selected should depend on the desired task and required information. Popp and Farber (1991) found that symbolic presentation of route guidance information was superior to other visual presentation modes, including text and maps. Symbolic presentation had the lowest driver workload rating and best traffic safety behavior. Depending on the requirements of the system under design, the inclusion of both a graphic display format (instead of a map) and textual lists displayed in different situations may provide the most usable overall system (Lunenfeld, 1990).
If a textual direction list format is used, it is critical that in–vehicle information can be received in short glances, so as not to distract the driver from the driving task. The recommended optimal message length for text format is less than eight words. Vague terms should be avoided and nomenclature should be tested (Dingus and Hulse, 1993). Messages should be standardized to enhance familiarity of messages and shorten retrieval time. Systems should be able to give information both in terms of cross streets or landmarks and distances to incidents or destinations.
Less attention is required in a well–designed turn–by–turn visual display than in a full–route format. Little information is required for a graphic turn–by–turn screen, including direction of turn, distance to turn, and turn–street name. This information is easily displayed in a legible, low attention–demanding format. Any additional information may be extraneous and potentially disruptive to the route–following task (McGranaghan, Mark, and Gould, 1987). Therefore, turn–by–turn graphic displays are recommended over full–route maps in most circumstances.
In selected circumstances, a more complex screen graphic may be required. One such circumstance, is that of close–proximity maneuvers. Many circumstances exist in the driving environment for which two (or more) quick turns are required. In the turn–by–turn format case, the information for the second turn may come up too quickly (and under circumstances where attention is needed for driving) to comfortably execute the second maneuver. Therefore, a graphic depiction of all maneuvers within a certain time envelope should be displayed.
Selecting either a turn–by–turn or route visual display requires careful consideration to ensure that the information is displayed in a usable and safe manner. If a turn–by–turn configuration is used, close proximity and preview must be considered; and if a full–route map is used, it is necessary to minimize the information present so that drivers are not overloaded. Even with the minimization of full–route map information, it is not clear from the literature whether or not drivers will become overloaded in high attention–demanding circumstances (Dingus and Hulse, 1993).
The north vs. heading–up display is another issue inherent in route–map presentation, especially if the speed and accuracy with which the display information can be interpreted by the driver is important (Dingus and Hulse, 1993). With a north–up orientation, the driver must often mentally rotate the map image (e.g., if the heading is southeast) to determine whether to turn right or left. This operation requires additional attention and processing time and results in more errors for the population as a whole.
One advantage to north–up map presentations is that they do not "move." For a heading–up format, the map must constantly rotate to maintain heading up as the vehicle heading changes. This rotation, particularly presented in the visual periphery, can be somewhat distracting, although most drivers seem to be able to ignore it (i.e., at least to some extent) if required to do so by the driving situation (Hulse, 1988).
According to Dingus and Hulse (1993), heading up is a better option than the north–up trade–off. However, they emphasized that both methods have disadvantages and recommended that an appropriately designed verbal direction list or turn–by–turn spatial option is more desirable than a route–map using either presentation rule.
When developing advanced information systems, the problem of "out–of–the–loop" loss of familiarity needs to be considered (Dingus and Hulse, 1993). Presently, the driver is required to obtain most information from the driving environment (i.e., street signs, stop lights, etc.). As more information is presented within the car in a readily accessible manner, the less likely the driver will need to obtain that information from the driving environment. However, any problem, deficiency, or inconsistency that requires the driver to shift attention to the driving environment could result in a delay and increased effort since the driver has become accustomed to having the information provided within the car. Therefore, it is important that all information provided in–vehicle be accurate and reliable. If the accuracy of critical information cannot be reasonably guaranteed, then that information should not be provided.
Information provided to the driver should be timely and sufficient time must be allowed for the driver to respond to it. The driver needs time to hear and/or see the information, decide whether it is relevant, and act upon it. More than an average human response time is required; most drivers (i.e., 95 percent or 99 percent) should have ample time to respond under normal driving conditions. The time required by the driver to process information and respond to it is dependent upon a number of factors, including the task and the type of display format selected. This information is beyond the scope of this paper. A discussion of driver response time requirements can be found in several sources, including the NHTSA Driver Performance Data Book (1986).
Visual display considerations
A delineation of appropriate display parameter options (e.g., resolution, luminance, contrast, color, glare protection) is a complex topic that will be somewhat task and situation specific. There are a number of legibility standards that exist for visual displays, including those developed for aircraft applications (Boff and Lincoln, 1988). Since automobiles and aircraft have many of the same difficulties, (e.g., glare), these standards are applicable. However, note that the selection of a display in the automotive domain will be more constrained by cost and have limitations well below the state of the art.
In any event, display parameters have a minimum acceptable level and anything below that level is unusable. It is therefore critical to ensure that these minimum standards have been met in spite of the constraints for a given application. In addition, the problem of selecting automobile display parameters is more difficult, since viewing distance is limited due to instrument panel configuration constraints.
The user population also has difficulties in determining display parameters. Older drivers with poorer visual acuity and/or bifocal lenses must be carefully considered when specifying display parameters. To overcome this combination of limitations, the display parameters must be optimized within practical control. For example, Carpenter et al. (1991) used a "special high–legibility font" in a color CRT application to ensure that drivers could glance at the display and grasp the required information as quickly as possible. Other design aspects that aid in legibility include presenting text information in an upright orientation (even in map applications), maximizing luminance and/or color contrast under all circumstances, avoiding selected color combinations (Boff and Lincoln, 1988), using color–coding of information sparingly (since too many colors create more information density and increase search time), maximizing line widths to increase luminance (particularly on map displays), and minimizing the amount of displayed information to reduce search time (Dingus and Hulse, 1993).
Color deficiency and color blindness is another basic visual display concern. Approximately 8 percent of the male population has some degree of color deficiency or color blindness. Therefore, it is important not to color–code critical information in consumer product applications (including navigation systems).
Although instrument panel color has been shown to have no significant effect on reading and driving performance (Imbeau et al., 1989), Brockman (1991) found that color on a computer display screen can be distracting if used improperly. To avoid confusion, Brockman recommended several guidelines when color–coding information. The first guideline was to be consistent in the use of color codes. Designers should avoid using colors from extreme ends of the color spectrum (i.e., red and blue) next to each other since it is difficult for the reader's eye to perceive a straight line. Second, familiar color coding (e.g., red for hot) should be used. Third, color alone should not be relied upon to discriminate between items. Brockman recommended designing applications first in black and white, then adding color to provide additional information.
The placement of an information display is critical. The information contained on even a well–designed display system requires a large amount of visual attention. Therefore, if the display is placed far from the normal driving forward field of view, none of the driver's peripheral vision can be used to detect unexpected movement in front of the vehicle. Another disadvantage of placing a display far away from the forward field of view is increased switching time. The farther the display is away from the roadway, the longer the switching takes, and the less time that can be devoted to the roadway or the display (Weintraub, Haines, and Randle, 1985). Green and Williams (1992) found that navigation displays were recognized at a faster rate when using a HUD over a dash–mounted display. Tarriere, Hartmann, Sfez, Chaput, and Petit–Poilvert (1988) reviewed some ergonomic principles of designing the in–vehicle environment and suggested that the screen be mounted at least 15 degrees below horizontal, but should not exceed 30 degrees, for optimal driver comfort.
A HUD that provides information on the windshield is a good choice since it is in the forward field of view. HUD's are also advantageous because they focus at (or near) infinity, thus eliminating (or reducing) the time required for the driver's eyes to adjust between the display and the roadway. However, a number of concerns have been raised by Dingus and Hulse (1993) about the use of HUD's:
Therefore, the design of a HUD must be carefully developed and tested.
Manual control guidelines
A system that requires too many of a driver's resources may not leave enough resources to allow the driver to drive safely in all circumstances. This means that both navigation displays and navigation system controls need to be designed to minimize driver attention and processing of resource requirements.
One way to minimize control requirements while in transit is to severely limit control access. Therefore, it is important to assess the necessity of every control in terms of both in–transit requirements and frequency of in–transit use to minimize driver control access. Those controls that are not absolutely necessary for the in–transit environment can then be allocated to pre–drive or zero–speed circumstances.
Several other specific control issues are also present with respect to ATIS applications. First, control location is important since the farther away a control is located from the driver, the greater the number of resources needed to activate the control. This has been demonstrated by Bhise, Forbes, and Farber (1986) and Mourant, Herman, and Moussa–Hamouda (1980) who found that the probability of looking at a control increased with increased distance. Therefore, controls present on the steering wheel, or otherwise in close proximity to the driver, are easier to use.
Complexity is a second important control factor to consider. Research (Monty, 1984) has shown that continuous controls or controls requiring multiple activations are significantly more difficult to operate. Therefore, limiting controls to single, discrete activations will provide fewer resource requirements.
A third control factor is the trade–off between "hard" and "soft" push buttons. With the use of CRT's and flat–panel displays, there has been a strong temptation to use touchscreen overlays for control activation. While this can be a good control method in the automobile for pre–drive or zero–speed cases, this is not the case for in–transit circumstances. Monty (1984) found that the use of touchscreen keys while driving required greater visual glance time and resulted in greater driving and system task errors than conventional "hard" buttons. The reasons for this performance problem are twofold:
For a "hard" button, the driver must (depending on the control and its proximity) glance briefly at the control and then find the control using tactile information to accomplish location "fine–tuning." For the soft keys, the driver must glance once to determine the location and glance again to perform the location "fine–tuning." Therefore, while soft keys are often convenient, they require greater driver resources and are not recommended for navigation information system in–transit application.
Providing information to convince drivers to avoid congested routes
Several papers (e.g., Allen et al., 1991b, 1991c) suggested guidelines to help improve route selection and route diversion. Advisory systems should provide anticipated traffic congestion information. Current conditions can be viewed as unstable and subject to rapid change. Advisory systems should target information that allows commuters to change their departure time and route choice. This information should be accurate and visually confirmed by the use of road signs or on–board congestion monitors. Information should give projected commute times of both alternative and present route choices.
Design guidelines to accommodate age and alcohol effects
ATIS design should include alcohol–impaired drivers and drivers from all age groups. Both alcohol–impaired and older drivers need advanced roadway hazard warning systems due to their slower reaction times. These drivers would also benefit from advanced signing of upcoming exits. Information systems should use icons to improve user visibility and increase character size of textual labels to improve performance problems experienced by elderly drivers. Older drivers have shown that they respond better visually to the use of yellows, oranges, yellow–greens, and whites on contrasting backgrounds. Moving pointers are also preferred over digital or number displays.