市區道路網絡中以全球定位系統資料為基礎之速度模式預測與路徑指引

Abstract

交通擁塞在市區而言是個重要的問題，它會使得駕駛們浪費龐大的時間與金錢。近年來裝載有GPS定位裝置的車輛逐漸普及，而這些車輛的位置資訊對於估計複雜的市區道路網絡的交通狀況相當有用，根據準確的交通狀況估計，我們可以提供適當的行駛路徑給駕駛們，便能夠避開擁塞路段，減少時間與金錢的浪費。 這篇論文中，我們利用行駛於市區道路網絡中的車輛之GPS位置來估計路段的交通狀況，我們提出了一個速度模式模型來描述路段上的交通狀況，並且也提出一個分類基礎的路徑指示模型，透過機器學習技術來學習歷史交通狀況的演變。路徑指示模型將能夠依據駕駛所處的交通狀況及它由歷史資料所學習到的經驗來決定適當的路徑提供給駕駛們參考。

Traffic congestion is an important problem in city. It could lead a significant waste of money and time. In recent years, cars equipped with GPS devices become widespread and the location information of those cars could be very useful to estimate traffic condition in the complex city road network. According to the accurate traffic condition estimation, we can provide appropriate route guidance to road drivers and they can avoid the congestion. In this thesis, we use the GPS coordinates of cars driving on the city road network to estimate the traffic condition of road segments. We propose a speed pattern model to describe traffic condition as the travel speed pattern. And we propose a classification-based route guidance model by learning the historic traffic data using machine learning technique. The route guidance model could provide route guidance to drivers according to current traffic condition and how traffic condition would change by the experience learned from historic traffic data.