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
Guidelines From Human Factors Guideline Documents
When developing new systems, experimental testing of every design aspect is expensive, time–consuming, and unnecessary. Rather than testing every feature separately, engineers rely on guidelines, principles, and "rules of thumb" to aid in their design decisions.
Existing applicable human factors guidelines for developing ATIS/CVO systems will be discussed in detail in the following paragraphs.
This section will not cover those areas where human factors guidelines, such as information processing, are not given or are vague. Although information processing contributes a useful model of human cognition, research does not contain many "cut–and–dried" facts that are useful for ATIS/CVO designers.
Anthropometry is defined by Kantowitz and Sorkin (1983) as "the application of scientific physical measurement methods to the human body in order to optimize the interface between humans and machines and other manufactured products." People come in a wide variety of sizes and shapes. Therefore, designers must try to accommodate the widest range of human physical dimensions possible in their designs. By doing so, they will maximize the number of people who can use the equipment.
There are three strategies given by Kantowitz and Sorkin (1983) for using anthropometric data in design:
When designing an ATIS system, anthropometric data can be helpful in providing guidelines for three important considerations.
First, manual controls should be located so that all drivers can easily reach them. One anthropometric measurement, the arm reach envelope, provides information on how far people can reach with their hands in an entire range of vertical and horizontal directions. Specific parameters for this type of measurement can be found in NASA's Man–Systems Integration Standards (1989).
Second, visual displays should be placed so that they are easily readable for all drivers. In NHTSA's Driver Performance Data Book (1986), studies of drivers' eye positions were measured in relation to the car–body inch lines. The distribution of eye positions for the 2,300 subjects sampled was represented as a series of concentric ellipses that can be used to determine what drivers can and cannot see from their eye positions.
Fortunately, seating is adjustable to some extent for most automobiles and commercial vehicles. This feature might eliminate the need for adjusting the position of some ATIS controls and displays.
Third, controls should not be placed so that they are inadvertently activated by the driver's natural body position. For example, a driver's right knee might lean against the dashboard and disturb controls. Also, visual displays should not be located where they will be blocked by the user's body.
Human–Computer Interaction (HCI)
In many respects, an ATIS/CVO system is similar to a typical desktop computer. Designers of the HCI interface for an ATIS system need to design an interface for a heterogeneous population with varying skills, experience, and appreciation for computers.
The following is a general list of HCI guidelines by Williges, Williges, and Elkerton (1987).
Compatibility. Language used in a software interface should have clear and unambiguous meanings to users. It is important that the meaning of the language be compatible with user population stereotypes, given the diversity of language use and the imprecision of language application by humans.
Controls and displays should also be highly compatible. An example of stimulus–response compatibility is a HUD signaling a driver to turn right using an arrow located on the right side of the display, as well as pointing to the right. Research has also shown that auditory input and speech responses are more compatible with verbal tasks, and that visual input and manual responses are more compatible with spatial tasks. This suggests that speech controls might be beneficial in an ATIS/CVO system when they become economically and technologically feasible. The research also shows that visual displays are more compatible with manual responses. However, this research must also be reconciled with studies showing that a driver's visual capacity is already taxed by the driving task, and that further demands on the visual channel could have safety risks.
Consistency. User input and system output should be consistent between screens and software modules. For example, the help button on a touchscreen display should be in the same position, regardless of the current functions of the screen.
Memory. In the design of human–computer dialogues (i.e., question and answer, menu selection), it is important to minimize the amount of information users must store in short–term memory, especially if other information processing is simultaneously required. The maximum number of items a person can remember is between five and nine. However, this number also depends on the complexity of the items, the sequence of presentation, the length of time they must be remembered, and the amount of competing information to be processed. Since using an ATIS system will often be a secondary task (driving being the primary task), there will often be a significant amount of competing information to be processed. Therefore, the demands on memory should be minimized.
Structure. Providing structure to the system's functions will help a user form an internal representation of the system. For example, by describing a word processor as analogous to a typewriter, users are able to use their existing knowledge of a typewriter in order to help them understand the functions of a word processor. Moreover, by providing users with the system structure (e.g., a hierarchical tree of software functions and modules), the user will be able to perform better in unfamiliar sections of the software. In ATIS/CVO systems, giving users a tutorial hierarchical overview might help them acquire knowledge.
Feedback. Whenever users make a system input, such as requesting traffic information, the system should give a response or feedback to users to acknowledge receipt of this input. If the system cannot give the information back in a timely manner, it should advise users when the output will be given. By doing this, users realize that the input has been received, and they understand that the delay in the system response is not an error on their part. A system should also supply feedback if the input made was an error (e.g., a distinctive "beep" if users try to select a more detailed route map when none exists). It should also inform users why the input was an error and suggest alternative choices.
Workload. Because users' performances suffer if the mental workload of a system is too high, every effort should be made to minimize the complexity of the ATIS/CVO system, especially the density of the information displayed, in order to reduce workload. Furthermore, extra information should be removed if not needed. These recommendations suggests that users should have control over the rate of information displayed, especially if it is auditory information, since it cannot be scanned or ignored.
For a more extensive list of HCI guidelines, designers should consider Smith and Mosier's Guidelines for Designing User Interface Software (1986). This report gives 944 HCI guidelines, many with examples, comments, conditions of applicability, and cross–references to other guidelines. As stated by the authors, the goals of these guidelines are to provide consistency, minimize user input, minimize user memory requirements, maximize compatibility, and maximize user flexibility with the computer. These guidelines are broken down into six categories summarized below. Note that several of the categories apply to ATIS/CVO systems.
Data entry. Data entry refers to user actions involving computer input and computer responses to this input. Most of these guidelines are not applicable to ATIS systems due to the limited nature of data input necessary for ATIS use. An example of a guideline from this section is entitled "Storing Frequently Used Text: Allow users to label and store frequently used text segments, and later to recall stored segments identified by their assigned labels." This guideline might, for example, prompt designers to allow all destinations that an operator has previously keyed in (e.g., "454 Main Street") to be accessed by a menu number, instead of having the operator retype the entire destination.
Data display. Data display refers to computer output to a user and assimilation of information from such outputs. This category includes guidelines on map displays–a valuable resource when developing a navigational (IRANS) system. However, many of the listed guidelines were derived from full–size CRT screens and are not applicable to the smaller screens found in vehicles. An example of a guideline from this section is entitled "Aiding Distance Judgments: When a user must judge distances accurately on a map or other graphic display, provide computer aids for that judgment." Using this guideline, designers might put mileage scales on ATIS maps.
Sequence control. Sequence control refers to user actions and computer logic that initiate, interrupt, or terminate transactions. An example of a guideline from this section is entitled "Logical Ordering of Menu Options: List displayed menu options in a logical order; if no logical structure is apparent, then display the options in order of their expected frequency of use." Such a guideline might prompt designers to place an "emergency roadside service" option after other more frequently used options, such as "display restaurants/motels."
User guidance. User guidance refers to error messages, alarms, prompts, and labels, as well as more formal instructional material provided to help users interact with the computer. An example of a guideline from this section is entitled "Task–Oriented Help: Tailor the response to a help request to the task contents and current transaction." This guideline might prompt designers to offer specific destination information in a navigational task, if requested during the task.
Data transmission. Transmission refers to computer–mediated communications among system users, and also with other systems. This category is applicable to vehicle–to–infrastructure and infrastructure–to–vehicle communications. An example from this category is entitled "Automatic Feedback: Provide automatic feedback for data transmission confirming that messages have been sent or indicating transmission failures, as necessary, to permit effective user participation in message handling." Designers might use this guideline to provide feedback on whether an emergency roadside assistance message has been sent, or if there is a problem with transmitting the message.
Data protection. Protection is necessary for security from unauthorized use, potential loss from equipment failure, and user errors. Guidelines for this subcategory may be more applicable to CVO systems where sensitive business information might require more data protection than an individual's travel itinerary. An example of a data protection guideline is entitled "User Confirmation of Destructive Actions: Require users to take an explicit extra action to CONFIRM a potentially destructive control entry before it is accepted by the computer for execution." The designer of an ATIS system might use this guideline to require users to confirm that they want to delete a destination entry in a navigational task.
Another resource for HCI guidelines is the Military Standard (Mil–Std) 1472D (1989, pp. 247–278), which provides 30 pages of guidelines, although it does not give any conditions of applicability or cross–references. Mil–Std 1472D covers only the critical and general guidelines. An example of a guideline from this report is shown below:
"Page numbering: Each page of a multiple–page display shall be labeled to identify the currently displayed page and the total number of pages, e.g., page 2 of 5."
ATIS designers could use this guideline for numbering pages of on–line help text.
Another article that deals exclusively with linguistic guidelines for HCI is an article by Harris (1990), which provides a list of 10 guidelines for proper use of the English language in a computer interface. Many of these guidelines are similar to those mentioned previously. However, the article is informative.
General display issues
The most common types of ATIS displays use the visual and auditory modalities to display information. The sections below look at each of these two modalities separately, as well as compare and contrast their benefits in an ATIS/CVO system. Finally, tactile displays are briefly discussed.
Visual vs. Auditory Displays. To decide whether information should be displayed visually or aurally, a designer must consider the user, the system, and the environment. Sorkin (1987) presented a list of conditions and considerations to aid in this decision:
Use auditory presentation if:
Use visual presentation if:
Sorkin also pointed out that the omni–directional nature of auditory displays makes them suitable for alerting and warning messages, such as for IVSAWS applications.
There is significant evidence to support the use of auditory displays in vehicles in addition to or instead of visual displays. First, many authors have found that giving turn–by–turn directions on the auditory channel leads to quicker travel times, fewer wrong turns, lower workload, and more attention given to the primary task of driving than does route information displayed on a visual map (Labiale, 1990; Parkes, Ashby, and Fairclough, 1991; Streeter, Vitello, and Wonsiewicz, 1985; McKnight and McKnight, 1992). Second, there is evidence to support the notion that driving performance suffers when drivers are simultaneously looking at in–vehicle displays. Drivers tend not to react to situations on the road (McKnight and McKnight, 1992), and they deviate from their course (Zwahlen and DeBald, 1986).
On the other hand, there is also support for use of visual displays in navigational tasks. Aretz (1991) stated that if drivers are not presented with some sort of north–up map, they cannot construct an internal cognitive map of their route. This means that drivers might arrive at a destination with no navigational problems, but not have any idea where their destination is located in relation to their route origins or other landmarks.
According to Williges, Williges, and Elkerton (1987), another problem is that for a spatial task such as navigating, performance is optimal when using visual stimuli and manual responses. If the task is verbal, an auditory stimulus should be coupled with a verbal response from the person. These results suggest that for a navigational task, a visually displayed "arrow" indicating travel direction would be more beneficial than encoding the information verbally, as in an auditory speech display.
The guidelines presented here were constructed with several goals in mind. First, users should be able to identify visual stimuli. Second, they should be able to discriminate among visual stimuli. Third, the visual stimuli should not be contaminated with other "noise" in the environment.
The guideline's scope primarily includes the physical nature of the stimuli, rather than the cognitive aspects. For cognitive considerations, see the section on "Human–Computer Interaction." Furthermore, many guidelines concerning mechanical gauges and other mechanical displays are not included here because ATIS systems have to display many different types of data. As a result, it is not practical to use dashboard space for fixed–function mechanical displays.
The following visual display guidelines are given by Helander (1987).
Distance Between Displays – The larger the distance between displays, the slower the reaction time, and the more errors a user will make. Therefore, in–vehicle displays should be located close to the front windshield for optimal driving and ATIS performance.
Grouping Display Information – Display information that should be considered simultaneously by a driver should be presented in accordance with the Gestalt principles of proximity, similarity, continuity, and closure. For example, in an ATIS navigation screen, the Gestalt principles of proximity would suggest that origin information and destination information should be physically separated on a screen by at least one or two blank lines in order to avoid confusion over whether information should be categorized as origin or destination.
Contrast Ratio – This is the ratio of the object luminance over the background luminance. A high contrast ratio will result in better perception. One of the more accurate and popular measures of this is the Modular Transfer Function Area (MTFA). For a high–contrast display, an MTFA value of 10 or greater is recommended. For more information on this measure and its calculation, see Snyder (1985).
Character Resolution – For high screen resolution, a dot matrix of 5x7 pixels is required, but 7x11 and 9x11 matrices are preferred. In addition, because square pixel dots take up more space, they are a better choice than round dots.
Font – Character legibility depends on font. Basically, the closer the font approximates regular stroke characters, the higher the legibility. For 5x7 pixel matrix characters, the Huddleston font is the best, and for 7x9 and 9x11 matrix sizes, the Huddleston and Lincoln/Mitre fonts are equally good. Woodson (1981) recommended that fonts that avoid character confusion be selected. For example, the letters C and G can be confused if the horizontal stroke in the G is too short.
Font Size – Researchers recommend a size of 14 to 22 minutes of arc, with 18 minutes being an optimum size for reading. For visual search tasks, such as looking for a street name on a map, 22 or 24 minutes is recommended. Note that measurements are given in terms of the observer's visual angle, since this corresponds to the retinal image and eliminates the need to give both object size and distance from object measurements.
Screen Reflections – There are many sources of screen reflections in a vehicle, including the sun, headlights from another vehicle, and reflections from another vehicle. In order to reduce screen reflections, a tiltable screen or screen filters should be used. Another technique is to display the characters in reverse video–dark characters on a white background. However, reverse video may cause problems with screen flicker, since the illuminated area is so large. For a detailed list of measures to reduce screen reflection and their advantages and disadvantages, see page 529 of Helander's study (1987).
Discriminating Among Colors – Observers should be able to discriminate among colors on a CRT. Formulas to determine discrimination performance are based on color saturation, color hue, and the contrast ratios of different colors. These formulas can be found on pages 538–539 of Helander's report (1987).
Coding – No more than five colors should be used on a display for high–accuracy identification. Colors should be used to convey information–not for aesthetic value. In an ATIS map display, categories of roads, such as highways or secondary roads, should consistently be given a specific color.
Uses of Color – Color is useful if operators must search for information. For example, in a display of upcoming restaurants along a particular route, those restaurants with take–out service could be shown in a particular color.
Conventional Meanings of Color – Research suggests that red, yellow, and green be reserved for "danger," "caution," and "safe," respectively. IVSAWS warnings could be coded in this manner for dangerous environmental conditions, such as ice or snow, and for heavy traffic caused by accidents.
Colors in Text – In alphanumeric displays, red and yellow colors against a black background are the most legible, while blue and green are the least legible.
Color and Object Size – Smaller objects lead to poorer color recognition than larger objects. Therefore, color–coding is more effective for entire words than for individual characters or small symbols.
The following alerting signal guidelines are from the report by Boff and Lincoln (1988).
Dual–Modality Alert – Present high–priority alerting signals both visually and aurally. Maximize the probability of detection of each mode of the warning signal.
Location of Alert – Place visual alerting signals as close to the operator's line of sight as possible. The maximum deviation of 15 degrees should be allowed for high–priority alerts and 30 degrees for low–priority alerts.
Size of Alert – Visual alerting signals should subtend at least 1 degree of visual angle.
Contrast Ratio – Visual signals should be at least twice as bright as the other displays.
Flashing – Visual alerting signals should be flashing against a steady–state background.
False Signals – False signals should be minimized, and a method of canceling the signal should be provided.
Note that the source of alerting signal information is originally from an aircraft study, and should be approached with caution. For example, slight deviations from an aircraft's direction of travel is rarely hazardous, unlike ground–based transportation. Therefore, an alert that startles operators so they temporarily go off course for a minute is an acceptable risk in an airplane, but not in a car. For a more in–depth look at visual display image quality and visual information portrayal, designers should consult pages 2216–2310 of the report by Boff and Lincoln (1988).
As mentioned earlier, there are many advantages to using auditory displays rather than visual displays. However, there are also more potential problems with noise interfering with the auditory channel than with the visual channel. Typically, designers only have to worry about the "noise" of screen reflections for visual displays. Unfortunately, just as auditory signals are omni–directional, noises and competing signals in the system or in the environment are also omni–directional.
Designers of an ATIS/CVO system must be particularly aware of the sounds in an automobile or a commercial vehicle. Road noise, noise from one's own vehicle, noise from other vehicles, competing speech communication from passengers in the vehicle, competing signals from the vehicle's entertainment system (e.g., radio, compact disc (CD) player), and other communications equipment (e.g., cellular radio, CB radio) must be taken into account when designing in–vehicle auditory displays.
The following guidelines are given by Sorkin (1987).
Signal Levels – Signal levels of 15 to 16 dB above masking threshold are sufficient for situations requiring a rapid response to a signal (e.g., a warning signal). (Note: masking threshold is defined as the sound level required for 75 percent correct detection of a signal when presented to the observer in a two–interval task. In a two–interval task, the observer reports which of two defined observation intervals randomly contains the signal. Both contain noise.)
Maximum Signal Levels – The level of an auditory signal should be less than 30 dB above masking threshold in order to minimize operator annoyance and disruption of communication.
Minimum Duration – The minimum duration signal burst should be at least 100 ms to ensure reliable detection, but not much longer, since other communication could be disrupted.
Pitch – The pitch of warning sounds should be between 150 and 1000 Hz.
Signal Spectra – Signals with harmonically regular frequency components should be used instead of inharmonic components. For lower priority warning signals, most of their energy should be in the first five harmonics. For high–priority signals, more of the signal's energy should be in harmonics 6 through 10. These high–priority signals can also be made distinctive by incorporating a small number of inharmonic components.
Onset and Offset Rates – Since fast onset rates for the pulse shape of a warning tone may produce a potentially dangerous startle response from the operator, the onset and offset rates should be limited to 1 dB/ms.
For an in–depth look at auditory signal characteristics and perceived urgency, see the Edworthy, Loxley, and Dennis (1991) report.
A major problem with using only simple auditory signals is that operators must be able to not only distinguish between signals, but properly associate the particular signal with its meaning. Technology has permitted the development of artificially produced speech displays that are capable of communicating with an operator in his/her own language. Such displays allow for a much higher rate of information transmission than do simple auditory warnings.
Sorkin and Kantowitz (1987) discussed two general factors that determine the quality of speech. First is articulation, which is measured by how well individual syllables and phonemes are recognized. This is determined by how distinct the speech is from the noise. The second is its intelligibility–the comprehensibility of the words, sentences, or the total message. Intelligibility is determined by the size of the message set and the relative probability of a message being chosen from the set. More specifically, the larger the message set, the poorer the comprehension. Also, intelligibility is poorest when all messages have an equal probability of being presented.
It is important to note that understanding speech is not just a bottom–up process, but a top–down process as well. Measurement of speech articulation–sonic characteristics–is insufficient to determine how well a person will understand it. Comprehension is measured by intelligibility, where peoples' expectations (i.e., probability of a message being presented), as well as other cognitive factors, determine their understanding.
The following speech display guidelines are presented and referenced according to their source.
Signal–to–Noise Ratio – Speech intelligibility increases as the signal (speech)–power–to–noise–power increases (Boff and Lincoln, 1988).
Voice Type – Select a voice type according to the source of the speech messages. For machine messages to the operator, use machine–sounding voice quality; when simulating human speech, use human–sounding voice quality (Simpson, McCauley, Ronald, Ruth, and Williges, 1987). Bertone (1982) recommended using only a female voice, based on a military helicopter in which only males were flying. Therefore, the female voice might be more distinctive among male communication. However, in an ATIS/CVO system, gender distribution will be about equal and will eliminate the distinctiveness of a female voice.
Prosodics – Prosodics are the natural pitch undulations in human speech. Regardless of voice type, use the best approximation of natural prosodics (Simpson et al., 1987).
Rate of Speech – For warning messages, use a speaking rate of approximately 150 wpm. A slower rate may be desirable for training listeners who are unfamiliar with the speech accent. Pending further research, the best rate for a given application will have to be determined experimentally (Simpson et al., 1987).
Alerting Tones – When a machine–quality voice is used exclusively for warnings, do not put any alerting non–speech sound before the speech warning message. When a machine–quality voice is used for warnings and for other functions (e.g., advisories, responses to user queries, etc.), incorporate an alerting characteristic into the voice warnings. Possible alerting features may include higher voice pitch, alerting speech or non–speech prefixes, or other features that make the warning message distinctive and can be shown to increase detectability without increasing human–system response time (Simpson et al., 1987). Boff and Lincoln (1988) suggested that warning messages be prefaced with the operator's name.
Sound Location – When speech signal and masking speech are presented from different loudspeakers, signal intelligibility increases as the distance between signal and masking speech increases (Boff and Lincoln, 1988). Therefore, in an ATIS system, auditory information ideally should come from a unique location in the vehicle, far away from the possible locations of other passengers. This could be accomplished either by installing a special speaker or by having the sound come from a virtual location by using the present sound entertainment system in the vehicle.
Length – For warning messages, use a minimum of four syllables to provide sufficient linguistic context for warning comprehension after first enunciation of the message (Simpson et al., 1987). For increased intelligibility, sentences should be used instead of isolated words (Sorkin, 1987).
Content – Make message content appropriate for the task, and use terminology that is familiar to the users (Simpson et al., 1987).
Unreliable Information – Do not present unreliable information in the voice mode (Simpson et al., 1987).
Competing Voice Messages – Consider conflicts between multiple voice messages and between listening to voice messages and speaking (Simpson et al., 1987).
Priority of Messages – When delivering time–critical information by voice, as in warnings, incorporate a priority system to order concurrently triggered voice messages so that the most critical is presented first (Simpson et al., 1987).
Repeating of Messages – For warning messages, repeat the message after an appropriate time interval (see next guideline) only if the condition that triggered the warning message is still true (Simpson et al., 1987).
Duration Between Repeated Messages – The length of time before a warning message should be repeated depends on the severity of the consequences if the user does not correct the problem (Simpson, et al., 1987).
Spoken Menus – For spoken menus without concurrent visual display, limit the number of menu items to three (Simpson et al., 1987).
Peak Clipping – By clipping the positive and negative peaks of the speech wave, and then re–amplifying the remaining waves to the original amplitude, speech intelligibility can be increased in certain types of noise (Sorkin et al., 1987). Up to 20–dB peak clipping has no effect on intelligibility in the presence of "white" noise (Boff and Lincoln, 1988).
Filtering – Perfectly satisfactory speech communication is obtainable by filtering out all sounds below 800 Hz and all sounds above 2500 Hz (Sorkin et al., 1987). Filtering may be crucial when real–time information must be sent to vehicles, and where data transmission rates are often limited.
The tactile channel is rarely used as the primary channel to transmit information, but instead, is used as a redundant form of information. The most common use of the tactile channel is in the design of manual controls to provide feedback. On a computer keyboard, the "F" and "J" keys often have a raised surface to indicate the position of the index fingers on the home row. Many aircraft have a "stick shaker" tactile display connected to the control column. Whenever the plane is in danger of stalling, the control column will vibrate to alert the pilot to take corrective measures (Kantowitz et al., 1983). Tactile displays are also used on highways. Reflective strips placed on the white line of the emergency lane or on the yellow dividing lines not only enhance the road contours, but makes the car vibrate if the strip is crossed.
Unfortunately, no specific guidelines for tactile displays can be given. It can only be stated that an effort should be made to encode manual controls on an ATIS/CVO system with tactile information to enhance feedback and to allow drivers to manipulate controls without taking their eyes off the road.
ATIS/CVO systems do not require high volumes of data input by drivers. Drivers primarily select a menu or data item rather than input data. The philosophy of an ATIS system is that the machine (i.e., in–vehicle system and infrastructure) provides all of the information: route directions, maps, information on services and facilities, traffic warnings, and weather advisories. The primary control tasks of the user are to choose the information desired and to control its presentation.
Commercial vehicle systems might have higher volumes of data to input, due to the functions of an electronic credentials system and an electronic log book. Once again, however, the purpose of these systems is to eliminate the need for manually entering information such as mileage and location, since these data can be automatically recorded by the CVO system interfacing with the vehicle and with the infrastructure (i.e., fleet command).
The following guidelines are given for manual controls by the authors cited.
Task Selection – Use a keyboard if data entry volume is high (Greenstein and Arnaut, 1987).
Keyboard Layout – Use a standard layout. Even for untrained users, the QWERTY keyboard is no better or worse than an alphabetic layout (Greenstein and Arnaut, 1987).
Cursor Positioning Devices – Trackballs are the most accurate cursor positioning devices, although touchscreens and lightpens have led to faster response times (Greenstein and Arnaut, 1987).
Touchscreens – Touchscreens have many advantages:
Their disadvantages include:
One obvious disadvantage ignored by Greenstein and Arnaut is that touchscreens have no inherent feedback. It is crucial that supplementary feedback be given with touchscreen responses. Given these advantages and disadvantages, it seems that touchscreens can be recommended for an ATIS/CVO system, provided one need not enter high volumes of information, use this type of control for an extended period of time, or edit text on the screen. Note, however, that the lack of touchscreen feedback and tactile feel can be a problem in dual–task environments (as previously discussed). Therefore, touchscreens are only recommended for pre–drive circumstances.
Touchscreen Size – The screen should be large enough to accommodate enough sensors to allow for a one–key–per–letter keyboard. Therefore, a 5x6 or 6x7 matrix is recommended. If a smaller matrix is used, such as the 5x5 matrix currently used in the TravTek system, then a less efficient data entry form must be used, which leads to more errors and slower data entry (Dingus and Hulse, 1993).
Touchscreen Sensor Size – The sensors should not be smaller than 19 mm square according to the Mil Std 1472D (1989). However, anthropometric data should be consulted to determine the range of fingertip widths before selecting a final measurement.
Steering Column Controls – These types of controls should be safer than alternative controls (such as a touchscreen), since drivers' hands do not leave the wheel.
For an extensive listing of manual control guidelines, including physical measurements, see Woodson's report (1981).
Maintenance is often overlooked when designing systems. ATIS/CVO systems must be designed so that maintenance is easily accomplished. Mil Std 1472D (1989) provides an extensive list of design criteria for system maintenance. Listed below are some of the more applicable guidelines for ATIS systems.
Power Failure – Some indication should be provided when power failure occurs. All mission–essential electronic computer and peripheral system components should incorporate an automatic self–check software and hardware diagnostic program at power up and at the request of the operator to ensure that they are functioning properly. For example, when starting up a vehicle, the ATIS system should run a self–check on all functions, including communication functions such as external positioning systems (i.e., GPS) and cellular communications. Cellular communication is essential for roadside emergency assistance.
Out–of–Tolerance – A display should be provided to indicate when equipment has failed or is not operating within tolerance limits. An example would be a series of LED's inside an ATIS system chassis that indicate if certain elements are not working properly. Screen displays should also be used to display any malfunctioning aspects of the ATIS, such as cellular communications, CD–ROM, or hard drive. When a fault occurs, the system should provide the user with information on how to fix the system (e.g., take the vehicle to an authorized ATIS repair shop).
Printed Circuit Boards – Printed circuit boards should be designed and mounted for easy removal and replacement. Consider such factors as finger access, gripping aids, and resistance created by the mounting device. Appropriate feedback should be provided to ensure that technicians know when the board is securely connected.