An investigation was carried out to determine if a relationship could be found between occurrences in a traffic accident data base and visual demands created by various in-vehicle systems. The idea was to determine whether or not the assumption of a relationship between accident rate and visual demands is justified and if so, how to model the relationship. The advantage of modelling the relationship is that accident rates of proposed new or modified in-vehicle devices can then be predicted and in addition, problems associated with existing devices can be uncovered. This paper describes the investigative procedure used and the models that were developed. These models appear to have reasonably accurate predictive capability, given the necessary visual demand and frequency-of-use input data.
Abstract