Development of Human Factors Guidelines for Advanced Traveler Information Systems and Commercial Vehicle Operations: Definition and Prioritization of Research Studies
4. DISCUSSION
HIGHEST PRIORITY STUDIES/ISSUES
Table 19 lists the 9 most vital studies/issues followed by the 12 most important remaining studies/issues. Table 19 was built by combining List A (appendix B) with tables 16, 17 and 18. Items are not ranked within the two groups because differences within each group are quite small. These are the key issues that will guide future research.
Table 19. Most vital studies/issues.
| LABEL |
STUDY/ISSUE |
| A1 |
Examine the cognitive demands placed on the driver by the need to transition from one ATIS function to another. |
| A4 |
Identify how complex interactions among ATIS functions might affect driver understanding and response to the system. |
| B3 |
Identify how the information drivers need and want from road sign (ISIS) and warning systems (IVSAWS) might influence behavior. |
| C4 |
Examine how information reliability (e.g., false alarms) influences driver adaptation and enhance the potential for an improper
response to ISIS/IVSAWS. |
| C5 |
Investigate how to display multiple ISIS and IVSAWS messages so that drivers can identify relevant information and react
appropriately. |
| D2 |
Identify features that will benefit/require standardization across many types of ATIS systems and functions. |
| D4 |
Examine the performance differences associated with focusing all ISIS and IVSAWS information through either single or multiple
display channels. |
| D12 |
Evaluate the effectiveness of multimodality displays, such as voice in combination with text. |
| D17 |
Identify specific concerns regarding how display formats and modality impact CVO driver workload. |
| |
| B1 |
Identify how specific information needs vary as a function of driver characteristics (e.g., age, gender, etc.). |
| B2 |
Identify information drivers need and want from in–vehicle road signs (ISIS) and warning systems (IVSAWS). |
| B6 |
Describe the specific information needs/wants of CVO drivers for various situations, such as local versus long distance, urban
versus rural, and emergency response versus commercial. |
| C6 |
Examine how the timing of ISIS and IVSAWS information, with respect to the location of the incident, influences driver reaction
to the information. |
| D16 |
Assess how the fatigue that plagues CVO drivers (e.g., 8– to 12–h shift) might interact with complex in–vehicle
systems to degrade driver performance. |
| D20 |
Examine how information can be displayed to dispatchers to support the complex decision–making process associated with allocation
of emergency response crews. |
| D21 |
Examine how display design might aid the dynamic allocation of driver visual and cognitive resources. |
| E2 |
Evaluate how different types of information, displayed using a HUD, affect cognitive attention devoted to the roadway. |
| E10 |
Identify the compensatory actions drivers take to moderate their workload, and how ATIS/CVO systems might affect those actions. |
| F2 |
Examine how route guidance systems might adversely influence driver detection and recognition of unusual roadway events. |
| G1 |
Evaluate the effect of information reliability and inaccuracies on driver acceptance and use of ATIS/CVO systems. |
| J1 |
Investigate how to structure design guidelines to help designers address human factors issues in ATIS designs which support the
needs of the driver. |
BRIEF DESCRIPTIONS OF STUDIES/ISSUES
Tasks J and L will elaborate these issues in precise operational terms by developing workplans. To the extent possible these workplans will propose experiments that combine several issues in table 19 efficiently. There is not sufficient time to limit each experiment to a single issue. Thus, the following brief descriptions will be considerably expanded in Tasks J and L.
A1. Examine the cognitive demands placed on the driver by the need to transition from one ATIS function to another.
The Task C working paper discusses many ATIS/CVO functions and explains how certain subsets naturally are linked. Thus, at least two kinds of cognitive demands can be studied: those transitions that occur within a natural set of ATIS and those that cross set boundaries. Using a part-task simulator in a laboratory we can investigate effects in both classes of function transition. The simulator will provide objective measures of cognitive demands such as driver accuracy and latency. If necessary, results can be described by quantitative models such as the theory of signal detection and information theory.
A4. Identify how complex ATIS interactions among ATIS functions might affect driver understanding and response to the system.
This issue is closely related to A1 above and the same methodology can be used to study it. However, the main focus of interest is not transitions across functions but steady-state operation of one function that is related to other functions. This issue also includes potential problems associated with unexpected behavior of functions when a change in one function disrupts smooth operation of another function. It should be possible to combine issues labeled A1 and A4 into a single set of laboratory experiments.
B3. Identify how the information drivers need and want from road sign (ISIS) and warning systems (IVSAWS) might influence behavior.
This study will be performed on-road as part of Tasks L and M. Information that is presented to a driver must go through three mental stages. First, it needs to be perceived correctly. Second, a cognitive decision about how the information might alter the driver's goals must be formulated. Third, this mental plan must be translated into action, such as taking an alternate route. The on-road study will try to investigate the effects of ISIS and IVSAWS upon all three intervening stages.
C4. Examine how information reliability (e.g., false alarms) influences driver adaptation and enhance the potential for an improper response to ISIS/IVSAWS.
Systems often provide inaccurate information to drivers. False alarms in particular alter mental criteria for decision making. There have been documented cases where operators turned off system warning devices due to high false alarm rates. This is also related to issue G1, consumer acceptance of unreliable ATIS/CVO devices. A laboratory simulator will be used to control the reliability and accuracy of presented information. Both objective measures (e.g., driver performance) and subjective measures (e.g., trust in system) will be recorded.
C5. Investigate how to display multiple ISIS and IVSAWS messages so that drivers can identify relevant information and react appropriately.
Operators can often get lost in a system that presents too many messages simultaneously. This has been dramatically illustrated in nuclear power plants when a system fault causes an entire message-tile mosaic to illuminate. The "quiet dark cockpit" used in Boeing airplanes is one way transportation systems deal with this potential problem. For drivers the issue is whether the system should be quiet and dark or should it continually be updating the driver, even when he or she is not requesting information. The answer to this question may also depend upon driver demography. A driving simulator will be used to combine the driving task with a secondary ATIS task. Dependent variables will include both driving performance as well as comprehension of multiple messages.
D2. Identify features that will benefit/require standardization across many types of ATIS systems and functions.
This study is similar to research on control-display compatibility. It is best conducted in two phases. First, a survey will be used to measure driver population stereotypes. This may vary with driver demography. Then a laboratory simulation will be used to validate the stereotypes by confirming that popular mappings are used more effectively than unpopular mappings.
D4. Examine the performance differences associated with focusing all ISIS and IVSAWS information through either single or multiple display channels.
The concept of divided attention and its effects upon operators have been studied intensely since the seminal limited-channel model of Broadbent in 1958. The advantages and disadvantages of separate input channels depend upon the perceptual aspects of the inputs as well as its complexity. Even with simple inputs, such as dichotic pairs of digits, rate of presentation determines if ear by ear (channel by channel) is better than pair by pair (message by message). So we should not expect a simple answer to this question, especially when driver demography is also considered. This will be tested in a laboratory part-task simulator using both visual and auditory message presentation.
D12. Evaluate the effectiveness of multimodality displays, such as voice in combination with text.
This issue is closely related to issue D4 above and both will be tested in the same series of experiments. Redundancy usually improves message comprehension. But this may interact with the nature of the message (e.g., ISIS versus IVSAWS). A laboratory part-task simulator will be used to determine empirically what these tradeoffs are.
D17. Identify specific concerns regarding how display formats and modality impact CVO driver workload.
The CVO driver has special needs in addition to those of the private driver. For example, he might need warnings about low bridges and weight restrictions that would require mental calculations and planning. Some display formats and modalities might make such cognitive activities easier or more difficult. It is especially important to prevent such mental workload from impeding the safe operation of commercial vehicles. The Battelle Heavy Vehicle Simulator will be used to study this. This research is related to other section D issues in table 18 so that careful planning will be needed to coordinate studies. It might be prudent to delay D17 until some initial results with private vehicle drivers have been obtained.
CONCLUSIONS
Task I evaluated 1 year of research that produced 8 large reports with a total of over 2,600 pages. Filtering all these results to obtain a prioritized list of key research topics was not a simple task. Producing the final list required a psychometric analysis of 17,472 data points generated by 8 human factors experts. Analyzing a set of issues that has already been selected from a very large set of reports is technically difficult, because the selected issues form a narrow range of items: inappropriate items have already been weeded out by the selection process itself. Nevertheless, the validation study showed that this psychometric analysis was successful. Thus, we can have great confidence that the final prioritized list is well-suited to guide future research. The key issues listed here must be addressed if human factors as a discipline is to make a substantial technical contribution to the development of ATIS and CVO components of ITS.
FHWA-RD-96-177
|
