請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/90752
標題: | 基於長短期記憶神經網路之非號誌化交叉口車輛軌跡預測 Vehicle Trajectory Prediction at Unsignalized Intersections using Long Short-Term Memory Neural Network |
作者: | 吳菘權 Sung-Chuan Wu |
指導教授: | 許添本 Tien-Pen Hsu |
關鍵字: | 非號誌化交叉口,長短期記憶神經網路,車輛軌跡預測,深度學習,交通安全, unsignalized intersections,long short-term memory neural network,vehicle trajectory prediction,deep learning,traffic safety, |
出版年 : | 2023 |
學位: | 碩士 |
摘要: | 在過去十年間,台灣交通事故的傷亡人數呈現持續上升的趨勢,不論是在非號誌化交叉口內或是非號誌化交叉口附近,肇事事件的數量也逐年增加。近年來,國內外學者紛紛運用深度學習技術來預測車輛行駛行為,因此,本研究使用深度學習的方法,建立一個有效的神經網絡模型,用於預測非號誌化交叉口車輛的軌跡。
本研究以台北市八德路三段74巷與八德路三段12巷51弄、延吉街30巷的非號誌化交叉口作為研究交叉口,並進行了資料蒐集和調查。透過將蒐集到的影像轉化為軌跡資料,並訓練與驗證了基於長短期記憶神經網絡(LSTM)的模型。此模型不僅適用於所選研究地點,還可以泛化應用於其他類似非號誌化交叉口的車輛軌跡模擬研究。期望未來,此研究成果能透過提前警示方式,向鄰近非號誌化交叉口之駕駛者或用路人示警,進而減低交通衝突風險,提升交通安全。 In the past ten years, the number of casualties in traffic accidents in Taiwan has continued to rise. Whether in or near unsignalized intersections, the number of accidents has also increased year by year. In recent years, scholars have used deep learning technology to predict vehicle driving behavior. Therefore, this study uses the method of deep learning to establish an effective neural network model for predicting the trajectory of vehicles at unsignalized intersections. This study took Lane 74, Section 3, Bade Road, Taipei City, Lane 51, Lane 12, Section 3, Bade Road, and the Lane 30 Yanji Street as the research intersection, and collected and investigated the data. By converting the collected images into trajectory data, the model based on long-term and short-term memory neural network (LSTM) is trained and verified. This model is not only suitable for the selected research site, but can also be generalized and applied to other vehicle trajectory simulation studies which is similar to unsignalized intersections. It is expected that in the future, this research result can warn drivers or passers-by when they are currently near the unsignalized intersections through early warning, so as to reduce the risk of potential traffic conflicts and improve traffic safety. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/90752 |
DOI: | 10.6342/NTU202303869 |
全文授權: | 未授權 |
顯示於系所單位: | 土木工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-111-2.pdf 目前未授權公開取用 | 4.89 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。