本論文主要發展駕駛行為感知技術與行車風險管理系統， 利用車內診斷裝置(On-Board Diagnostic II，OBD-II dongle)，進行車載端訊號感測與蒐集，並透過感測數據分析，可即時判斷 駕駛行為是否出現異常，例如急加速、急煞車、急轉彎或怠速過久等異常行為，此外亦結合所開 發之智慧行車輔助APP (Driving Angle)，提供駕駛異常行為警示訊息，用以改善行車過程中可能 產生的不良行為， 提升駕駛行車安全性。同時事件資訊與原始數據(Raw data)將回傳至雲端平台 服務系統，透過所蒐集的車輛動態資料，進行駕駛行為分析、建模與驗測，建立駕駛風險估算機 制。此外建模精準度主要仰賴實車資料的長期蒐集，故場域試煉規模的擴大將有助於事件可信度 的提升，未來亦可串聯國內外產險業者，整合傳統風險精算機制與駕駛感知分析技術，發展下世 代金融科技。

In this study, the driving behavior sensing techniques and driving risk management system are provided. The on-board diagnostic II device (OBD-II) is utilized to collect the in-vehicle dynamic signals. The abnormal driving behavior, such as rapid acceleration, rapid braking, sharp turns or idle waiting, can be detected by our designed abnormal driving algorithm. Moreover, the auxiliary APP (driving Angle) is developed as a user interface to show the warning message. It would help the driver improve the driving behavior, and enhance the safety of driving environment. Meanwhile, the event information and raw data (Raw data) will be feedback to the back-end service platform. Finally, the driving risk estimation mechanism is established by analyzing the vehicle dynamics data. Moreover, the high accuracy of the model construction depends on crowd vehicle dynamic data collection. Therefore, the field trial scale needs to be expanded to enhance the event credibility. Finally, the property and casualty insurance industry should be jointed to develop the next generation financial technology by integrating the traditional actuarial risk p erception and driving sensing and analyzing technology.