Traffic research

Workload assessment is of interest to many applied settings ranging from VDT (visual display terminal) data-typing to space travel. Each area has its own specific problems. Much research was and still is performed in the area of aviation, in which, in particular, military flight studies have been carried out. In these studies pilots have to perform very complex tasks under extreme conditions and the selection of pilots is so stringent that in general only healthy young men and women are capable of realising these tasks. This selected group of people also serves as subject in aviation workload studies, where extreme G-forces and heart rate values of up to 160 bpm are no exception (see, e.g., Roscoe, 1993). These types of environments have a clear influence on physiological measures (e.g., on HRV quality, see Jorna, 1993).
No such forces are encountered in ground travel, though it could be argued that the human perceptual information system is unfit for the high speeds of travel possible in modern cars on motorways. Actual practice shows that this is not true and that many people are able to perform this task daily without negative consequences. Sometimes, however, our information processing system reaches its limit and things go wrong. In most of those cases a human error has occurred (Smiley & Brookhuis, 1987), resulting in a traffic-law violation that led to an accident (Rothengatter, 1991). In that case driving speed may have turned out to be too high to deal with safely. The selection criteria for driving are also far less strict than those applied in (military) flight, and, as a result, the population behind the wheel is far more diverse in capabilities. These factors, among others, make workload research in traffic an area with its own specific problems. Results booked in this area of research may benefit a large group of people.

Most workload measures used in traffic research have been developed, and tested, in the laboratory and in other applied settings such as the workplace (see e.g., Meijman & O'Hanlon, 1984) and aviation. The exceptions to these are the primary-task measures, for driving; the vehicle parameters.
A useful experimental design in traffic research is to compare task performance in an experimental (e.g., mental load) condition with performance under baseline conditions. A difference in performance can then be attributed to the experimental manipulation. Recently, however, Brookhuis (1995a, 1995b) has proposed critical levels of performance for different primary-task measures. These critical values can be considered performance margins as discussed in chapter 4. The criteria are not workload redlines, since they indicate the point at which performance should be considered to be affected, and thus indicate a shift from the A to the B region. Most of the measures' critical levels have been linked to unsafe behaviour, e.g., a level at which the likelihood that the vehicle leaves the traffic lane increases to a major extent (see Brookhuis 1995ab). In the following evaluations the absolute criteria will be included.

Evaluation of workload measures on their characteristics in traffic research will mainly be restricted to work that my colleagues and myself have performed. Self-report measures, primary-task and secondary-task measures, and physiological parameters have been used in these field studies. From these studies, specific road sections or conditions were selected with increased task demands. The studies will be divided into two categories, studies that include an increase in complexity and studies in which driver state is affected. The first category can be further divided into two sub-categories, an increase in road complexity versus an increase in task complexity, i.e. the addition of a secondary task. Differential sensitivity of a selection of measures to mental load in relation to demand are of primary interest in the evaluations.

Selected sections or conditions
From one simulator and six field studies, experimental and baseline conditions or road segments (sections) were selected and the sensitivity of workload measures were compared between conditions and load categories. Sections were selected based upon expected effect of stressors or environment on workload, i.e. a selection based upon task demand. The following baseline and load conditions were selected:

Complexity studies - environment:
1. From the `weaving section study', a study performed on the A28-motorway, driving over the combined entrance/exit road-section was selected as experimental condition and compared with a baseline control section. In appendix 1 the load condition is referred to as `ACC 2' (section ZL in the Dutch report). The baseline control section was a road segment with no entrance or exit and is indicated in the appendix as section `CTR1' (C1 in the Dutch report). All subjects drove these sections two times, once without eye-movement registration equipment, once with the equipment mounted on their heads (indicated as `c' for CEMRE, Continuous Eye Movement Registration Equipment).
2. In the `noise barrier study' the same eye-movement equipment was used in one condition. Driving over a road section near a noise barrier was used as experimental condition and compared with driving along the same road section in the opposite direction, where no such barrier was present. The mid-part of the noise barrier was far closer to the motorway than the begin and end part. Driving on the motorway along the barrier created the impression that the screen `approached' the car.
3. In the `road layout study' the effect of a changed road design of an `A'-class road on driving behaviour, in particular on speed choice, and on mental load was tested. The baseline consisted of driving on an ordinary A-road section that either preceded or followed an experimental section. All roads had a speed limit of 80 km/h and were single carriageways with two lanes, separated by a white line. The experiment included roads in two environments: a road leading through a forest (Wr, woodland road) and a road leading through open moorland (Mr, moorland road).

   Complexity studies - additional task:

4. The `car-phone study' was carried out both on a quiet motorway and on a busy four-lane ringroad. In the experimental conditions the drivers had to perform a difficult memory task, the PASAT, Paced Serial Addition Task (Gronwall & Sampson, 1974), while operating either a hand-held or a hands-free telephone set. In the experiment subjects drove and handled the car-phone five days a week for a total of three weeks. For the present comparisons, only data collected during the first week in which driver workload is likely to have been highest, were used.
5. The `tutoring' or DETER (Detection, Enforcement and Tutoring for Error Reduction) study is the only simulator study included in the comparisons. In this study, drivers had to complete four trials in a driving simulator where they drove through built-up areas, on A roads and on dual carriageways. The middle two trials, where an enforcement and tutoring system provided the subjects with feedback about detected violations, were compared with the first and last trial, when no feedback about violations was given. The tutoring messages and the required behavioural adaptation were suspected of increasing mental load.

   If baseline performance in the above-mentioned studies is assumed to be in region A2, performance in the load condition with increased demand can be expected to be mainly situated in the A3 region (see figure 2), and perhaps in the neighbouring left-hand section of the B-region (Table 3). In some of the conditions, in particular the conditions that included the use of the CEMRE, mental load may have been additionally increased. The CEMRE reduced the visual field and subjects were therefore required to make additional head movements (see also appendix 1). However, in none of the studies is demand expected to be excessive.
   Two studies in which driver state was affected were added. In the three load-conditions of the two studies, task difficulty was increased because driver state that was non-optimal as a result of the use of alcohol, a sedative drug or fatigue that followed lengthy driving. Performance in the load condition of these studies is expected to be, on average, situated in the A1 or D region (figure 2). The actual, individual region of performance, however, will depend upon individual capacity, experience and goals set for performance.

   Driver state studies:

6. In the `DREAM' (Driver Related Evaluation And Monitoring) study the effects of legally-allowed levels of Blood Alcohol Concentration (BAC <= 0.5 ‰), and the effects of fatigue (2.5 hours of driving, indicated in the article as `vigilance'), were compared with baseline performance (the first hour of the last-mentioned condition). Driving on a busy four-lane ringroad and on a monotonous motorway were included in the study.
7. Finally, in the `antihistamine study' the effects of a new-generation antihistamine (Ebastine) were compared with placebo and an active-drug control, Triprolidine. The active drug, which has a sedative effect, was chosen as the experimental condition and its effects were compared with the effects of placebo. In both conditions subjects had to drive on a busy four-lane ringroad and on a four-lane motorway.

Back to my HomePage