合理的運転者の定量的ヒューマンエラーモデルとその解析

概要

論文の詳細を見る
This paper extends a driver model called "maximum acceptable risk model" to consider explicitly a trade-off between accident reduction and travel time reduction under a situation which is either dangerous or safe. Three types of imperfect perception results of the situation are considered: "dangerous" denoted by "y", "safe" by "n" and "unknown" by "u". Results "y" and "n" may include premature alarm and lack of alarm, respectively. Parametric representations of conditional probabilities of perception results under a dangerous or a safe situation are given for rational drivers. Travel time reduction requires the driver to behave on a maximum acceptable risk level boundary. Conditional and unconditional probabilities of unnecessary avoidance are derived, their monotonic properties are clarified with respect to perception performance, and convergence to high-performance points along the boundary is shown. Accident occurrence probabilities and human error probabilities are obtained and decomposed into error types of perception, judgement and execution. Highly reliable accident prevention devices are useful not only for the accident reduction but also reduction of travel time.
社団法人精密工学会の論文
2006-08-05