ABSTRACT

Dynamic prediction of queues and delays at signalized intersections utilize data obtained from automated traffic sensors as inputs for Intelligent Transportation Systems (ITS) applications. Errors are inevitable during automated data handling, especially when the traffic involved is heterogeneous and lacking in lane discipline. This paper presents a model-based estimation scheme that can handle both mixed traffic conditions and erroneous detector input data to estimate queue and, in turn, delay. The models were developed for Queue within Advance Detector (QWAD) and Queue beyond Advance Detector (QBAD) scenario. The statistical properties of detector errors were incorporated into the estimation scheme, and the scheme was tested for varying traffic conditions. The estimation scheme’s performance was evaluated using field data from a signalized intersection in Chennai, India, and simulated data. It is found that the incorporation of statistical properties of detector error allows accurate estimation of queues and delays despite erroneous data input.

摘要

信号交叉口的队列和延误动态预测利用从自动交通传感器获得的数据作为智能交通系统 (ITS) 应用程序的输入。在自动化数据处理过程中，错误是不可避免的，尤其是当所涉及的流量是异构的并且缺乏车道规则时。本文提出了一种基于模型的估计方案，可以处理混合交通状况和错误的检测器输入数据来估计队列，进而估计延迟。这些模型是为高级检测器内的队列 (QWAD) 和高级检测器外的队列 (QBAD) 场景开发的。检测器误差的统计特性被纳入估计方案，并针对不同的交通条件对该方案进行了测试。使用印度钦奈信号交叉口的现场数据和模拟数据评估了估计方案的性能。发现检测器误差的统计特性的结合允许准确估计队列和延迟，尽管有错误的数据输入。