AbstractTravel time estimation and route suggestion play certain important roles in ITS, especially when applying to Advanced Traveler Information Services (ATIS). In recent years, the technique of Vehicle Ad-hoc Network (VANET) rises and develops quickly; VANET provides the capability of dynamic internet-connected communication as well as the possibility of exchanging real-time traffic information which can upgrade the efficiency of road network. This study supposes intersection traffic signals controllers are all equipped with roadside unit (RSU) which has VANET capability and vehicles are also all equipped with on-board unit (OBU). In city road network, this study will attempt to construct real-time travel time estimation and the shortest travel time route suggestion system of any intersection (point) to intersection (point).
The purposes of this study include: First, using data fusion scheme to construct travel time estimation model of city road network. The scheme would give appropriate weight to each data source to integrate data from different sources. It is expected that the uncertainty of data will decrease and the practicality of model will be improved, therefore more accurate and more reliable traffic information can be obtained. Second, by adopting the concept of VANET, real-time information communication between vehicle and intersection is feasible. It regards intersection traffic signals in city as intermediary roadside units. And then it calculates suggested route and estimates travel time by intersection RSU and along with the concept of collaborative computing, it exchanges and updates real-time traffic information between RSUs of intersections. It could not only alleviate the computing burden of back-end systems but provide drivers with the most reliable suggested route and travel time estimation value of each intersection. Therefore, this study will use data fusion to construct travel time estimation model and establish a travel time estimation structure of collaborative intersections. In this way a system of reliable real-time route suggestion and accurate travel time for drivers between any two intersections in city road network can be provided.
This study takes practical small city road network-sections in east and north district of Tainan City as research target. Data sources include the vehicle detectors (VD) data which is acquired from Tainan City Government Department of Transportation, and GPS report data from taxi fleet in Tainan (Chunghwa satellite fleet). Experiment results reveal that the accuracy of estimation cannot be maintained on a certain degree and has a higher degree of variation when only one data source is applied, either PV or VD. The travel time estimation becomes more accurate or maintains on a certain level when combining with real-time data and comparing with results adopting historical data only. The accuracy increases gradually or maintains on a certain level and the variation of model also decreases gradually when more step number of data fusion is adopted. As a result, under different temporal periods, most of the practical travel time of suggested route could save more time and maintain certain reliability than other routes.
|Date of Award||2011|
|Supervisor||Kevin-P Hwang (Supervisor) & Wei-Hsun Lee (Supervisor)|