The effects of visibility conditions per se are also apparent from Fig. 12.2. They were considerable, though hardly surprising.
The overall pattern in the results is that there appears to be a single CAS that brings down the amount of close following without generating an increase in average driving speed. This is the earlier "4 sec TTC + Added Counterforce" CAS. Somewhat surprisingly, the addition to this CAS of an extra constraint in the form of a 1-sec simple headway criterion worsened performance, both in terms of average driving speed and of the proportion of short headways that occurred.
The braking distance HUD showed no demonstrable beneficial effect on behavior, not even a reduction in short headways.
This study compared the behavioral effects of a number of reasonable CAS systems, that is, systems that constituted plausible ways of reducing the probability of rear-end collisions. There may, of course, be other plausible solutions and other reasonable parameter settings for the systems studied here. However, the present systems are themselves the more promising candidates from a larger set of systems (e.g., Janssen & Nilsson, 1990). For this reason, the results of the present experiment can be considered to have general implications.
The CAS systems evaluated here were found to differ substantially in their effects on driver behavior. Beneficial effects were consistently obtained only for the CAS that had already proven its worth in earlier work, the 4 sec TTC + Added Counterforce CAS. Compared to a control condition, this CAS reduced the amount of close following while not increasing overall driving speed. Thus, this CAS met the requirement of showing a beneficial effect on the primary car-following parameter that was not offset by counterproductive changes in another essential parameter. Also, the favorable effects associated with this CAS did not depend on the visibility conditions. The suspicion that the design of a well-functioning CAS should be differentiated according to external conditions therefore proved to be unwarranted.
It is interesting to speculate on the reasons, first, why there were significant differences between different forms of CAS support and, second, why there was no differential effect on performance of the supports in degraded visibility conditions. With regard to the first question, the failure to achieve favorable behavioral effects of both the HUD-like CAS and the CAS that used the combined 4 sec TTC + 1 sec headway remains to be explained. Post hoc, the HUD-like device may have failed exactly because it gives information in a form unrelated to the relative movement of the