bol-based systems, and it is possible that future systems should adopt a level of adaptation to cater for such individual preferences.
It must be noted that, although subjects may genuinely hold such opinions, there could be a novelty factor associated with such new technology. Longitudinal studies would overcome such effects but they are inevitably associated with high financial costs; indeed, it could be argued that a snap-shot study such as this has the advantage of allowing the worst case safety scenario (i.e., the initial use of a system) to be tested. However, it is likely that as the technology matures, such longitudinal studies may become more feasible.
A number of issues arose in the two studies that highlight the problems of evaluating first generation systems whose technology is, as yet, not 100% proven. First, several preferred routes had to be rejected for the studies, given differences in route selection algorithms between the systems, and errors with mapping data that resulted in the systems recommending drivers to make dangerous or illegal maneuvers. Second, insufficient processing speed meant that there was a lag in the moving map-based system and this caused a number of navigational errors, as discussed earlier. Finally, the tracking system for S1 was not totally reliable and, on a few occasions, the system would "drift" off the designated route, thus failing to provide guidance. This necessitated the rejection of some data in Field Study 2.
The trials revealed a number of deficiencies in the interfaces to all three route guidance systems. For instance, with the moving map-based system the lag and rotating aspects of the display meant that driver uncertainty was sometimes high and a number of errors subsequently occurred. With the symbol-based systems, the representation of distance to a maneuver caused some problems and, although recommendations for change can be made, further work is required to ascertain the optimum means of representing such information.
The approach adopted by S1 (i.e., simple standard symbols together with distance and street name/number) was found to produce a number of difficulties for subjects. However, such a technical solution lends itself to the presentation of information on a simple, low-cost display and so may be desirable for system designers/manufacturers. In addition, good human factors practice, such as the use of aerial view symbols ( Green & Williams, 1992) and the integration of distance countdown bars within the road layout representation, can enhance the usability of standard symbols. Therefore, if such an approach is to be adopted, careful consideration of the various components of the interface should take place to ensure their