Severval factors are critical in designing effective simulator systems for driving research and driver assesment. First, there is cost, in terms of which, less is better. Costly systems are sometimes fine for specific research purposes, but they are impractical for most other purposes. Second, there is realism; in this case, more may be better, but we don't know how much realism is really necessary for good prediction of actual driving behaviors. Recent evidence suggests that real-world clutter may enchance prediction, but it also makes interpreting research more difficult. More realism may increase cost, but this is not necessarily the case. Third, there is user-friendliness, in terms of which again, more is better. Systems that are easier to operate require less highly trained personnel; hence, increasing user friendliness also decrease cost, especially for assessment. We demonstrated a highly user friendly, low cost, moderately realistic, and partly interactive computer video driving research and assessment system. In recent research, we obtained multiple R values of .47 for older drivers with only three to four scenarios predicting a real world crash index.
Abstract