過去針對大客車之駕駛風格辨識，大多需要如：三軸加速規、六軸加速規、影像辨識等多種資料源進行整合計算，業者需投入大量軟硬體建設成本，使得國內大客車業者，在沒有政府補助下望而卻步不願投入成本建置，針對駕駛員不當駕駛行為之監控與管理，多透過事後調閱行車紀錄器的影片以及統計檢討；由於獨特的駕駛員人格特性，駕駛員具有不同的駕駛風格，此種被動式的檢核方式無法全面性地分析駕駛者行為，雖少部分廠商導入以GPS之速度作為急加速、急減速等危險指標，但可靠度過低。因此，駕駛風格將是駕駛員性能和安全駕駛能力可靠的衡量指標。
本研究利用國內大客車業者之真實數據，提出駕駛風格定義以及駕駛風格預測模型，以改善車機之事件偵測偽陽性過高之缺陷。本研究首先利用車機所記錄之數據以及行駛影像數據，依實際行駛情況之驗證及標記，並利用其車機事件以及時間區間，進行駕駛風格之條件定義實驗，證實後標記其駕駛風格，實驗結果顯示，本研究所提出之三項激進駕駛風格序列，與其相對應之ADAS車機事件相比，最多分別降低31.8%、54.5%以及65.5%，而其一疲勞駕駛風格序列則可降低約14.7%偽陽性。奠基於前階段所標記之駕駛風格資料做為真值數據，串聯車輛動態數據資料後，透過CNN（Convolutional Neural Network）、LSTM（Long Short-Term Memory）、GRU（Gate Recurrent Unit）以及SimpleRNN，提出一個僅需使用GPS之經度、緯度等少量數據，即可完成駕駛風格辨識之模型，實驗結果顯示以GRU具有最佳分類能力，各分類達7成以上精確度，並期望可提供大客車業者做為主動管理評估之指標。

In the past, for the driving style of buses, most of the needs such as: six-axis acceleration regulations and other data sources for integration. Operators need to invest a lot of construction costs, so bus operators do not want to invest in the cost of installation equipment. Because of the driver has different driving style, this passive way of checking cannot comprehensively analyze the driver's behavior. Although a small number of manufacturers import GPS speed as a rapid acceleration, but reliable is low. Therefore, driving style will be a reliable measure of driver performance and safe driving ability.
This study uses the GPS data and the driving image data, verifies and labels according to the actual driving situation, and uses the vehicle event and time interval to define the conditions of driving style. The experimental results show that the three aggressive driving style sequences proposed by this study were 31.8%, 54.5% and 65.5% lower than the corresponding ADAS vehicle events, while one Fatigue driving style sequence was reduced by about 14.7% false-positive. Based on the driving style data labeled as groundtruth data, through CNN, LSTM,GRU and SimpleRNN, proposed a model that can be used with a small amount of data to complete the driving style identification. The experimental results show that GRU model has the best classification ability, each classification reaches more than 70% accuracy, and it is expected that the bus operator can be used as an indicator of management assessment.

交通部. (2019). 智慧運輸系統中長期發展規劃.
衛生福利部統計處. (2019). 108年度死因統計.
Alkinani, M. H., Khan, W. Z., & Arshad, Q. (2020). Detecting Human Driver Inattentive and Aggressive Driving Behavior Using Deep Learning: Recent Advances, Requirements and Open Challenges. IEEE Access, Access, IEEE, 8, 105008-105030.
Consiglio, W., Driscoll, P., Witte, M., & Berg, W. P. (2003). Effect of cellular telephone conversations and other potential interference on reaction time in a braking response. Accident Analysis and Prevention, 35(4), 495-500.
Dorr, D., Grabengiesser, D., & Gauterin, F. (2014). Online driving style recognition using fuzzy logic. In (pp. 1021-1026): IEEE.
Eboli, L., Mazzulla, G., & Pungillo, G. (2017). How drivers’ characteristics can affect driving style. Transportation Research Procedia, 27, 945-952.
Ed-doughmi, Y., & Idrissi, N. (2019). Driver Fatigue Detection using Recurrent Neural Networks.
GuanHong, L. (2019). A Hybrid Deep Learning Framework for Driving Style Prediction Based on Vehicle Dynamic Data.
Lanjewar, R. B., Mathurkar, S., & Patel, N. (2015). Implementation and Comparison of Speech Emotion Recognition System Using Gaussian Mixture Model (GMM) and K- Nearest Neighbor (K-NN) Techniques. Procedia Computer Science, 49, 50-57.
Liu, H., Taniguchi, T., Tanaka, Y., Takenaka, K., & Bando, T. (2017). Visualization of Driving Behavior Based on Hidden Feature Extraction by Using Deep Learning. IEEE Transactions on Intelligent Transportation Systems, 18(9), 2477-2489.
Marina Martinez, C., Heucke, M., Wang, F., Gao, B., & Cao, D. (2018). Driving Style Recognition for Intelligent Vehicle Control and Advanced Driver Assistance: A Survey. IEEE Transactions on Intelligent Transportation Systems, Intelligent Transportation Systems, IEEE Transactions on, IEEE Trans. Intell. Transport. Syst., 19(3), 666-676.
Moukafih, Y., Hafidi, H., & Ghogho, M. (2019, 3-5 July 2019). Aggressive Driving Detection Using Deep Learning-based Time Series Classification. Paper presented at the 2019 IEEE International Symposium on INnovations in Intelligent SysTems and Applications (INISTA).
Osman, O. A., Hajij, M., Karbalaieali, S., & Ishak, S. (2019). A hierarchical machine learning classification approach for secondary task identification from observed driving behavior data. Accident Analysis and Prevention, 123, 274-281.
SU JIAN, C. J., HE JIE. (2012). Principal component analysis method and its application. Light Industry Science and Technology / Guangxi Journal of Light Industry(9), 12.
Xie, J., Hilal, A. R., & Kulic, D. (2018). Driving Maneuver Classification: A Comparison of Feature Extraction Methods. IEEE Sensors Journal, 18(12), 4777-4784.
Yan, F., Liu, M., Ding, C., Wang, Y., & Yan, L. (2019). Driving Style Recognition Based on Electroencephalography Data From a Simulated Driving Experiment. Frontiers in Psychology, 10, 1254.
Yang, L., Ma, R., Zhang, H. M., Guan, W., & Jiang, S. (2018). Driving behavior recognition using EEG data from a simulated car-following experiment. Accident Analysis and Prevention, 116, 30-40.