Experimental Studies on Driver's Route Choice Behaviour in Actual Road Network

本研究以实际驾驶员在路网中的路径选择行为为研究对象。从理论和观察两个方面对交通网络路径选择行为的研究进行了综述。研究发现，实际网络中的实验数据非常少，路径选择模型在实际网络中没有得到充分的验证。因此，在松山的实际路网中进行了路径选择实验。通过问卷调查和车载调查，观察出行前路线选择和实际路线选择，包括路线转换。驾驶员路径选择数据分析总结如下。假设驾驶员出行前的路径选择行为分为两个阶段：路径集生成过程和路径选择过程。在第一阶段，我们提出了一种提取起点和目的地之间的路径集的算法。将基于第k条最短路径算法的筛选方法与EBA(Aspects By Aspects)方法相结合。可以从实际的道路网络中提取可行路径集。通过与Dial的有效路径算法的比较研究，验证了该算法的有效性。可行集的路径选择阶段可以用多Logit模型来描述。我们加入了一些变量来解释网络的复杂性。它们是交叉口的结构和沿线道路功能的差异。研究发现，这些变量应该包括在路径选择模型中以及出行的时间和成本变量中。

This study focused on the actual driver's route choice behaviour in road networks. Previous studies on route choice behaviour in transport networks were reviewed both theoretical side and observation side. It was found that experimental data in actual networks were very few and route choice models were not sufficiently verified in actual networks. Thus, route choice experiment was made in the actual road network in Matsuyama. Pre-trip route choice and actual route choice including en-route switching were observed through questionnaire survey and onboard survey. Analysis of route choice data of drivers were summarized as follows. Pre-trip route choice behaviour of a driver is assumed to separated into two phases ; path set generation process and choice process. In the first phase, we have proposed an algorithm of extracting a path set between a pair of origin and destination. The screening method using the k-th shortest path algorithm was combined with the EBA(Elimination by Aspects) method. A feasible path set can be extracted from an actual road network. The validity of the proposed algorithm was verified though the comparative study with the Dial's efficient path algorithm. The route choice phase from a feasible set can be described using multiple logit model. We incorporated some variables explaining the complexity of a network. They are the structure of an intersection and the difference of road function along the route. It was found those variables should be involved into route choice model as well as time and cost variables of travel.