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-96-147
Date: October 1997
Development of Human Factors Guidelines for Advanced Traveler Information Systems and Commerical Vehicle Operations: Components of the Intelligent Transportation Systems: Designs Alternatives for In-Vehicle Information Displays
CHAPTER 1. INTRODUCTION
Advanced Traveler Information Systems (ATIS) are intended to provide travelers with real–time information on traffic and roadway conditions, vehicle navigation, roadway hazards, weather conditions, and motorist services. In–Vehicle Signing Information Systems (ISIS) and In–Vehicle Safety Advisory and Warning Systems (IVSAWS) are key components of the broader ATIS program, and have the potential to provide drivers with a wide variety of information via an in–vehicle display. ISIS is intended to provide drivers with information that is currently depicted on external roadway signs, such as non–commercial routing, warning, regulatory, and notification information (McCallum, Lee, Sanquist, & Wheeler, 1995). IVSAWS is intended to warn drivers of hazardous or unsafe conditions on the roadway ahead, including accidents, construction zones, and the presence of emergency vehicles (Erlichman, 1992).
While the technical capability exists to display ISIS and IVSAWS information in a variety of ways, little human factors research exists to guide the selection of a preferred display design from among the range of potential design alternatives. Key human factors considerations associated with selecting ATIS display alternatives include the accessibility, legibility, and understandability of ATIS information; the potential for ATIS information to facilitate driver decision–making; and the potential for ATIS information to distract the driver from the primary task of controlling the vehicle. A wide range of display design parameters are relevant to these human factors considerations. These design parameters must also be considered in the context of characteristics of the driving population and the environment. Driver characteristics include age and gender, and environmental characteristics include the existing ATIS and road–sign infrastructure. Figure 1 shows how driver attitudes and information–processing capabilities mediate the effect of ATIS design characteristics on driving safety and warning compliance. This figure also shows the range of variables selected for investigation. The general categories of ATIS design and driving context include only a few of the many possible variables. ATIS availability has been included as a design characteristic and as part of the driving context. ATIS availability can depend on design decisions and it may also depend on the roadway infrastructure.
In this study, we examine the effect of message style, the physical grouping or location of ATIS displays, the use of visual as opposed to auditory messages, the availability of the ISIS and IVSAWS information presented through an ATIS, and the availability of roadway information. This study will investigate how these factors affect driver compliance with warning messages, and whether particular ATIS message characteristics degrade driving safety.
One way to describe message style is by the degree to which it directs a driver to perform a specific action. At one extreme, messages might simply advise drivers of a particular roadway condition (e.g., "Icy Road Ahead"). At the other extreme, messages might command drivers to take specific actions in response to the condition (e.g., "Slow Down"). In the context of this discussion, we will refer to the endpoints of this dimension of ISIS and IVSAWS message design as the distinction between notification and command warning information. Notification information advises the driver of a particular condition, although the driver still maintains complete autonomy with respect to both interpreting the condition and deciding how to respond to it. Alternatively, command information suggests a course of action to the driver that if not followed, could adversely affect the driver. Giving information in the form of commands might require some automated integration of information. Such a system would evaluate existing roadway conditions, real–time traffic data, and vehicle data, and suggest an appropriate action to the human driver, although the driver is still free to either adopt or ignore the suggestion (see also Sheridan, 1982). While command information can reduce certain information–processing requirements of the driving task (e.g., perception of relevant information, integration of multiple sources of driving information, and complex decision–making), it also has the potential to misdirect drivers when the command fails to consider all the relevant factors, for example, a command to change lanes when the lane is occupied by another car.
Although message style has not been investigated empirically in the context of the driving task, studies in other domains have examined similar questions. For example, researchers have investigated the effects of mothers' message styles (directive or suggestive) on children's compliance. Suggestive styles have been found to lead to higher compliance in normal children (Crockenberg & Litman, 1990; Lytton, 1977; Rocissano, Slade, & Lynch, 1987), although the reverse is true for children with Down's Syndrome (Landry & Chapieski, 1989; Landry, Garner, Pirie, & Swank, 1994; Maurer & Sherrod, 1987). Extrapolating this result to the driving context suggests that notification information might induce greater compliance.
Research on the effect of directive and non–directive leadership style on performance and compliance is extensive. Generally, a directive style of leadership has been found to increase compliance (Lippitt, 1940), but reduce both group performance and satisfaction (Brollier, 1984; Hendrix & McNichols, 1982; Miller & Monge, 1986), especially when measured in field, as opposed to laboratory, studies. Simpler tasks (which are more common in the laboratory) are frequently performed better under directive leadership, while more complex tasks are best performed under a more democratic style (Becker & Baloff, 1969; Rudin, 1964). A less directive style seems to lead to better performance with highly anxious subjects (Misumi & Peterson, 1985). Extrapolating these results to the current study suggests that the command style would promote compliance, but that it might undermine driving safety and the driver's satisfaction and trust in the system.
The effectiveness of authoritarian leadership seems to be related to the amount of knowledge possessed by the leader. If the autocratic leader knows more than the other members of the team, authoritarian leadership leads to the most effective performance. If, however, the authoritarian leader knows less, or is given misinformation, the team will do much worse than if led by a less knowledgeable or misinformed democratic leader (Blyth, 1987; Cammalleri, Hendrick, Pittman, Blout, & Prather, 1973; Fiedler & Garcia, 1987; Shackleton, Bass, & Allison, 1975). This suggests a notification style for unreliable systems, or for messages that require interpretation and integration with other information sources.
Given the complexities and uncertainties associated with extrapolating these findings to ATIS design, it is difficult to draw firm conclusions. However, these findings suggest that a command style would promote greater compliance as compared to a notification message style. These benefits might be outweighed if drivers follow the ATIS commands blindly and ignore important roadway information. Thus, we might expect a trade–off for compliance and safety. This study will directly examine the differences in these two message styles and their effect on driver safety and warning compliance.
Another important dimension of ISIS and IVSAWS message design is the physical grouping or location of displays and displayed information. At one extreme, ISIS and IVSAWS information might be centralized on a single cathode ray tube (CRT) screen. At the other extreme, ISIS and IVSAWS information might be distributed across several locations using several display modalities (e.g., head–down instrument panel, center–mounted CRT, and head–up display (HUD)). For auditory displays, a centralized approach might refer to the concurrent use of all speakers in the vehicle, while a distributed approach might use individual speakers to localize sound and provide directional cues.
In the context of our discussion here, we will refer to this dimension of ISIS and IVSAWS message design as the distinction between centralized and distributed warning information. The literature on attention and time–sharing do not provide unequivocal guidance on the relative merits of a centralized versus a distributed approach. At issue is which approach to the display location makes the most effective use of available attentional resources and supports the most efficient time–sharing between assimilating warning information and the primary task of driving.
In this regard, the centralized display option may be a somewhat simpler and more parsimonious approach because it reduces driver requirements to attend and visually scan more than a single in–vehicle display. Specifically, attentional resources associated with instrument scanning can be focused on a single display, rather than "spread out" across numerous displays. There is some support for this approach in the literature. For example, Shaw (1984) reported that signal detection performance for simple stimuli decreases as the number of display locations increases. Similarly, Konrad, Kramer, and Watson (1994) compared multiple displays to a single sequential display using a simple monitoring task. They found that both response time and response accuracy performance were better with the single (centralized) display. Also, a centralized approach may address the tendency of people to engage in less–than–optimal scanning strategies when multiple displays must be monitored (Moray, 1981).
In contrast, a distributed display could use display location as an alternate code to the urgency or type of the displayed information. Thus, the location of the information can serve as a cue that may reduce processing time and aid driver interpretation, decision–making, and response associated with the information. For example, information requiring an immediate response from the driver can be presented on a HUD (which generally minimizes eye and head movement requirements), while less urgent information can be presented on one or more head–down CRTs. Although poor time–sharing will take place if the spatial separation across displays is too great (Wickens, 1984), the fixed and consistent nature of distributed displays may lead to parallel processing of visual information (Schneider & Shiffrin, 1977), resulting in automatic processing of the different information sources.
In sum, both the centralized and the distributed approach are associated with a number of theoretical and operational trade–offs. The centralized approach may be more effective with relatively simple stimuli and responses, and minimizes the number of displays that must be attended to by the driver. The distributed approach may be more effective with more complex tasks involving real–world decisions and demanding psychomotor responses. The distributed approach also uses display location as a redundant code to facilitate the interpretation of displayed information.
For many travel–related information displays (e.g., speed–limit signs and traffic signals), designers have little choice regarding the sensory modality used to convey the information. However, with ATIS information, designers have the option of using either auditory or visual displays to present ISIS and IVSAWS information. From the perspective of driver acceptance and performance, the most effective display modality is not always obvious and there are a number of design trade–offs and considerations associated with auditory and visual displays. The auditory channel can have an advantage over the visual channel due to its attention–getting qualities (McCormick & Sanders, 1982). In addition, auditory messages are an attractive option since driving already places high visual demands on drivers. The use of auditory messages might allow better time–sharing of limited processing resources, i.e., time–sharing between two sensory resources may be superior to sharing within a single resource (Wickens, 1984). However, the visual channel is the more traditional mode for the presentation of driving information, is associated with relatively higher information rates (Sorkin, 1987) than the auditory channel, and is less likely to startle the driver than auditory messages. While components of sounds such as speed, fundamental frequency, repetition units, and inharmonicity have been successfully manipulated to vary the perceived urgency of sounds (Hellier, Edworthy, & Dennis, 1993), relative urgency is but one component of ISIS and IVSAWS messages that auditory displays will need to communicate to drivers. The relative salience and urgency that can be conveyed by an auditory message may depend on message style and the dimension of directiveness. Command messages may be more compelling when delivered in an auditory, rather than a visual, format. Another concern with auditory displays is that they are frequently disabled by users due to an increase in frustration and subjective workload (King & Corso, 1993).
Although the selection of auditory versus visual displays depends upon a number of situation–specific variables, Deatherage (1972, p. 124) has provided general guidelines for selecting sensory modalities. These guidelines are presented in table 1. This table demonstrates the uncertainty regarding the most appropriate use of auditory and visual displays for ISIS and IVSAWS information. Either the exact nature of the ISIS/IVSAWS message is unclear (e.g., does it call for immediate action?), or the guideline is ambiguous (e.g., what if the message deals with both events in time and location in space?). Although table 1 does not provide a definitive answer, its guidelines suggest that an auditory display may be best suited to ISIS/IVSAWS information. This study will investigate this option directly.
Table 1. General guidelines for the selection of auditory versus visual forms of information presentation (Deatherage, 1972).
ATIS AND ROADWAY INFORMATION AVAILABILITY
Related to these display design issues is the availability of the ATIS and roadway information and the corresponding level of trust that the driver places in this information. ATIS information cannot be assumed to be consistently available, especially during initial implementation of ATIS. ATIS information may be inaccurate or non–existent. When ATIS information is not readily available, drivers may ignore the ATIS and use roadway or other information to guide their decisions. If driver trust in the system is low due to the lack of ATIS information associated with past experience, drivers may spend additional time verifying the accuracy of the information. In general, users are reluctant to rely upon equipment that they do not trust (Lee & Moray, 1992).
As ATIS becomes more widespread, drivers may encounter ATIS messages paired with redundant roadway information, roadway signs alone, and ATIS messages alone. In these situations, the driver must adapt his or her own knowledge of the roadway situation and available signage to the information provided by the ATIS and act accordingly. While chronic ATIS failures may lead drivers to disregard ATIS completely, intermittent failures will require drivers to adapt to the availability of ATIS information, following the ATIS when available and using roadway information otherwise. An ATIS that provides warning information intermittently may also undermine drivers' trust. This effect may depend on the message style. Command messages will be more useful to drivers and will be associated with higher levels of driver trust and acceptance when ATIS information is consistently available. With command messages, drivers are not given the underlying reason for the warning, while notification messages describe the situation and provide drivers with a better basis for making a decision if the information is unreliable. Thus, the command message style may be most appropriate for high levels of information availability, while the notification message style may be appropriate for lower levels of information availability.
Drivers using ATIS will be required to adapt their sampling of roadway and ATIS information as the availability of ATIS and roadway information varies. Trust in the ATIS and self–confidence are likely to be important factors mediating their sampling strategy (Lee & Moray, 1992). Factors such as the amount of driving experience and the costs associated with the missing information will affect this decision. For example, if the ATIS system does not display the speed limit of a section of highway that has been raised from 35 to 45 mi/h (56.3 to 72.4 km/h), the cost of the missing information is relatively low. However, if we reverse the situation, with the system not having the information that the speed limit has been reduced from 45 to 35 mi/h (72.4 to 56.3 km/h), the cost may be higher. That is, the driver may be driving at a speed that is unsafe for that section of highway and may risk getting a costly speeding ticket. This experiment will examine how drivers' trust in the ATIS interacts with their self–confidence to influence compliance with ATIS messages.
The redundancy of ATIS and roadway information is an important consideration for ATIS implementation. One option for ATIS implementation is to present ATIS messages without corresponding roadway messages such as changeable–message signs. Another option is to augment signs with redundant in–vehicle messages. Redundant information might enhance driver compliance with the warnings, but it might also overload the driver with too much information. The redundancy of ATIS and roadway information, and the associated information processing load, may depend on ATIS message style (notification versus command), display locations (centralized versus distributed), and sensory modality (auditory versus visual).
Considering different levels of ATIS and roadway infrastructure development, the three dimensions of ATIS display design generates four issues of particular importance: (1) the effect of notification versus command message style on compliance and driving performance, (2) the effect of centralized versus distributed display location on compliance and driving performance, (3) the effect of auditory versus visual ISIS and IVSAWS information on compliance and driving performance, and (4) the interaction of these issues in the context of potentially unavailable ATIS information and/or unavailable roadway information. ATIS information must facilitate rapid and accurate response; however, it must not degrade driving safety by interfering with drivers' ability to consider roadway information.
OBJECTIVES AND HYPOTHESES OF THIS STUDY
The overall objective of this study is to develop general guidelines for addressing ISIS/IVSAWS displays. This study examines three important dimensions of ISIS and IVSAWS design: message style (notification versus command), display location (centralized versus distributed), and sensory modality (auditory versus visual). In the context of inaccurate information from the roadway and the ATIS, the study examines how these dimensions affect drivers' ability to comprehend ISIS and IVSAWS information and combine it with roadway information to make appropriate decisions.
This experiment investigates how ATIS design characteristics combine with environmental and driver characteristics to influence compliance with warning messages and driving safety. In general, ATIS messages should enhance drivers' reaction to roadway events by encouraging greater speed reductions and faster lane changes. Furthermore, certain ATIS message formats may be more effective than others. While ATIS messages may encourage faster and more effective responses to roadway hazards, they may degrade driving safety. ATIS messages may degrade driving safety by overloading drivers. Alternatively, ATIS messages may encourage overreliance on the ATIS and draw a driver's attention into the vehicle and away from the roadway. The aim of this experiment is to identify the relationship between ATIS design characteristics, message compliance, and driving safety.
To examine the effect of different ATIS design characteristics on warning compliance and driving safety, several dependent measures were collected. In addition, this experiment examines several intervening variables to better understand the cognitive processes underlying the influence of ATIS designs on safety and compliance. Trust and self–confidence have been shown to underlie reliance on automation and information systems (Lee & Moray, 1992; Lee & Moray, 1994). Factors influencing trust and self–confidence are likely to influence compliance with ATIS warnings. Similarly, situational awareness, workload, and information assimilation measures should reflect the factors underlying how ATIS messages might compromise driving safety. Specifically, measures of workload should reflect the information overload that might accompany the additional information provided by the ATIS. Situational awareness and message acknowledgment should indicate whether ATIS draws attention away from the roadway. These considerations are reflected in a series of general hypotheses and a set of more specific hypotheses. The general hypotheses include:
The following specific hypotheses guided the analysis:
TRUST IN ATIS AND SELF–CONFIDENCE
SITUATIONAL AWARENESS, WORKLOAD, AND INFORMATION ASSIMILATION
Beyond these specific hypotheses, figure 2 provides a general framework for approaching the analyses. This figure places the dependent variables in a context that shows the relationship among dependent variables and between the independent and dependent variables. This figure highlights the four primary analyses. The hypotheses provide a starting point for each analysis and this figure provides a framework for considering the more exploratory data analyses.