圓環的混合車流特性及行為分析

Abstract

臺灣的機車持有率相當高，機車的機動性較高，容易進行超車、鑽車等行為，使得台灣市區道路上有複雜的混合車流現象。圓環路型漸漸受到歡迎，因圓環較一般交叉路口來得安全，且在臨界交通量下，能提供連續不斷的交通流動，又圓環對路口中央的古蹟而言有其存在的必要。一般圓環績效評估的指標可分成兩大面向，分別為容量與安全，這兩個面向底下都有一些特性需要被分析。又過去研究指出直線道路上，混合車流中機車的車流特性與汽車是有差異的，因此推測圓環上的混合車流與純汽車車流的特性也存在差異，但過去的圓環研究大部分都是針對純汽車車流，這些研究結果可能就不適用於機車持有率很高的地方。 為此，本研究的目的為提供一套以微觀車流軌跡資料分析混合車流於圓環上的特性與行為的流程，微觀車流軌跡資料的蒐集方式為先以無人飛機空拍圓環車流影片，將影片經過後製處理後，藉由軌跡擷取軟體以半自動方式蒐集車流軌跡資料，透過軌跡資料分析以下特性與行為：車道使用、速度、加減速度、橫向位置、變換車道次數、接受餘間距。分析上述特性與行為之前，以極座標判斷車輛於每個時間點的所在車道、以及建立一元二次曲線方程式估算車輛每秒行駛距離，有了上述基本資訊後，即可做後續分析，透過視覺化展現各個特性在圓環上的分布情形，另外也使用了圖表以及敘述性統計的方式呈現上述車流特性與行為的結果。 從研究結果發現機、汽車行駛於圓環上的特性與行為存在差異，而車輛的行駛特性與行為會受到道路彎曲程度、所在車道，以及當下為準備進圓環、在圓環內、準備出圓環的狀態影響。另外，從橫向位置的分析發現汽車在圓環上並無行駛在車道中央，推斷是因為車輛在圓環上經常變換車道所導致的，而變換車道的次數也會受到車輛繞行的距離(角度)以及行經多少個進出口所影響。接受餘間距的分析結果顯示各車種組合下的臨界餘間距存在差異。未來研究可將透過此流程分析圓環上的車流特性與行為，並更進一步建構混合車流在圓環上的微觀車流模式，最後再將上述結果應用於評估圓環績效或車流模擬軟體中。

The motorcycle ownership rate in Taiwan is high. Motorcycles are small in size, lighter weight and high mobility, and the characteristics are significantly different from those of passenger cars. Motorcyclists are more easily to do overtaking and filtering than cars. Roundabouts are widely used in urban areas. That could increase safety and operational performance, and is sometimes used in urban city to preserve a heritage. Generally, there are two indexes, safety and capacity, to evaluate the roundabout performance. Some traffic characteristics need to be analyzed before the study evaluates the roundabout performance. Previous study point out that the characteristics of mixed traffic and passenger car traffic on the urban arterials are different. This study infer that they are also different on roundabout. However, previous study relative to roundabout only focus on the passenger car traffic, the result may not be appropriate to apply to the high motorcycle ownership rate region. The objective of this study is to propose a process about how to use the microscopic traffic dataset to analyze traffic characteristics and behavior of mixed traffic flow for roundabout. This study adopted an aerial videography approach to collect the video using a drone. Videos are converted into microscopic traffic data in the form of vehicle trajectories, in which the characteristics and parameters of interested can be extracted from the dataset. The mixed traffic characteristics and behavior, such as lane usages, speed, acceleration, deceleration, lateral position, lane changing and lag acceptance are analyzed. The data extracted from the trajectory extraction software is in Cartesian coordinate system. In order to determine which lane the vehicle used at each data point, the computation is more convenient if the positions and trajectories of vehicles are in Polar coordinate system. The vehicles move along a curve in a roundabout, and it is not appropriate to use the straight line distance between two consecutive data points to represent the distance travelled. A quadratic function is used to fit in the coordinates of the data points before and after the time instances, and the distance travelled in a time interval is estimated by integrating the function over the time interval. The preliminary results show that there are significant differences in terms of lane choices, speed distribution and frequency of lane changing times between motorcycles and passenger cars at different locations of a roundabout. This study also found that the car drivers didn’t abide by the lane discipline on roundabout because the drivers change the lane frequently. The number of lane changing times would affect by the travel distance and how many entrance or exit the vehicle passed, and are different from cars and motorcycles. The result of critical lag show that there are differences under different combination of objective vehicle mode and conflict vehicle mode. Future research could follow this process to study the characteristics and behavior of mixed traffic for other roundabout. The next step is to investigate the interactions between vehicles, such as overtaking and filtering behavior of motorcycles (Wong and Lee, 2015). The derived characteristics and interactions can be further used for the development, calibration and validation of behavioral models of vehicles in a roundabout. The results can also be applied to the evaluation of roundabout performances for capacity and safety.