A Study on Evaluation of Traffic Information Considering Interference between Divers under Uncertainty

本研究提出了一种测量感知出行时间变异性、感知出行时间带宽的方法。对日本大阪地区高速公路用户进行的一项为期六周的日记调查结果显示，感知带宽比记录的出行时间的标准差更能解释安全裕度的变化，对可变性的感知在解释出行决策时很重要，客观的可变性度量，如出行时间的标准差，可能不是解释不确定情况下旅行者决策的有效变量。对来自76名受试者的实验数据的分析，支持了一些基于代表性出行时间感知的假设。研究表明，最近可获得性适用于习惯化之前对具有代表性的旅行时间的感知。研究结果还表明，习惯化后，被感知的代表性旅行时间与经历的旅行时间之间的关联性减弱。通过对实验数据的分析，验证了极值方法和锚定方法的有效性。本研究的数值算例表明，即使提供了精确的行程时间信息，网络流量的改善也可以忽略不计。在不存在预测行程时间信息的情况下，基于智能体的路径选择仿真结果表明，驾驶员的认知能力水平越高，驾驶员的自组织性越强，网络流量越稳定。它还表明，关于司机没有选择的路线的信息对于这种自我组织是重要的。仿真结果还表明，预测行程时间信息即使是准确的，也是无效的。

Proposed in this study is a measure of the perception of travel time variability, perceived travel time bandwidth. Results of a six-week diary survey of expressway users in the Osaka metropolitan area of Japan have revealed that perceived bandwidth explains the variation in safety margin more than does the standard deviation of recorded travel times, that the perception of variability is important in explaining travel decision, and that an objective measure of variability, such as the standard deviation of travel times, may not he an effective variable to account fur traveler' decision under uncertainty.The analyses of experimental data, obtained from 76 subjects, lend support to some of the hypotheses postulated on the perception of representative travel time. It has shown that availability by recentness applies in the perception of representative travel time before habitualization. The results have also shown that the association between perceived representative travel time and experienced travel times weakens after habitualization. Neither availability by extremity nor anchoring has been demonstrated by the analysis of the experimental data.The numerical examples of this study provide results that improvements in network flow are negligibly small even when precise information on travel time is provided. In case where no predictive travel time information is present, results of agent-based route choice simulation indicate that higher levels of drivers' cognitive capabilities tend to produce more self-organization of drivers and stable network flows. It is also shown that information on routes that a driver did not take is important for this self-orgarnization. Simulation results also indicate that predictive travel time information is ineffective even when it is accurate.