摘要 駕駛者必須付出連續且高程度的專注力在駕駛行為上。警覺性降低抑或心智負荷過高都會導致人為失誤的發生。而人為錯誤往往都是交通事故的主因。因此,瞭解駕駛者的心智負荷是非常重要的。 在這篇論文中,模擬如同平常高速公路的虛擬實境。利用不同狀況的環境或是非預期事件的發生來影響駕駛者的心智負荷。同時,駕駛績效、生理指標、主觀評量值會被擷取下來當成重要資訊。 根據駕駛指標、生理績效與主觀評量值的資料,利用多元迴歸與多項式網路來建構預測心智負荷的模式。在多元迴歸模式中,主觀評量值與平均車速、煞車平均變動量以及心跳有明顯的相關性。此研究的結果可用來發展適合的輔助系統並且潛在地提升交通舒適與安全。
Abstract Driving task consumes a great deal of operator attention continuously. Either low vigilance or information overload may lead to human errors. Human errors were always major cause of traffic accidents. Therefore, understanding operator mental state is important. In this study, the virtual environment of freeway was simulated where the drivers drove as usual. Mental workload of bus drivers were affected via different conditions or unexpected events. At the same time, driving performance, physiological index, and subjective ratings were measured during or after driving. We constructed a multiple regression and polynomial neural networks to predict mental workload which are evaluated by data from subjective ratings, task performance, and physiological indexes. In multiple regression model, it is found the mental workload is effectively related to average speed, average braking depth variation, and heart rate (p=0.000<0.01). The results in this study can be referred to develop adaptive aiding systems and also are potential to enhance comfort and safety in traffic.