The introduction of autonomous vehicles (AVs) in the near future will have a significant impact on road traffic. AVs may have advantages in efficiency and convenience, but safety can be compromised in mixed operations of manual vehicles and AVs. To deal with the issues associated with mixed traffic and to avoid its negative effects, a special purpose lane reserved for AVs can be proposed to segregate AVs from manual vehicles. In this research, we analyze the effect on efficiency and safety of AVs in mixed traffic and in a situation where an AV-only lane is deployed. In the analysis, we investigate the average speed, the throughput, and the inverse time-to-collision (ITTC). We differentiate the behaviors of manual vehicles and AVs through the reaction time, desired speed, and car-following models. As a result, we observe that the efficiency is improved when the market penetration rate of AVs increases, especially when the highway throughput increases by up to 84% in the case of mixed traffic. However, safety worsens when the market penetration of AVs is under 40%. In this case, the average speed can be improved and the frequency of dangerous situations (ITTC > 0.49) can be reduced drastically in the merging section by making the innermost lane AV-only. Accordingly, we conclude that AV-only lanes can have a significant positive impact on efficiency and safety when the market penetration rate of AVs is low.

Travel time is valuable information for both drivers and traffic managers. While properly estimating the travel time of a single road section, an issue arises when multiple traffic streams exist. In highways, this usually occurs at the upstream of diverge bottleneck. The aim of this paper is to provide a new framework for travel time estimation of a diverging traffic stream using timestamp data only. While providing the framework, the main focus of this paper is on performing a few analyses on the stage of travel time data classification in the proposed framework. Three sequential steps with a few statistical approaches are provided in this stage: detection of data divergence, classification of divergent data, and outlier filtering. First, a divergence detection index (DDI) of data has been developed, and the analysis results show that this new index is useful in finding the threshold of determining data divergence. Second, three different methods are tested in terms of properly classifying the divergent data. It is found that our modified method based on the approach used by Korea Expressway Corporation shows superior performance. Third, a polynomial regression-based method is used for outlier filtering, and this shows reasonable performance even at a relatively low market penetration rate (MPR) of probe vehicles. Then, the overall performance of the travel time estimation framework is tested, and this test demonstrates that the proposed framework can show improved performance in distinctively estimating the travel times of two different traffic streams in the same road section.

No abstract

Various mobility services have been proposed based on the integration of automated vehicle (AV) and road infrastructure. Service providers need to identify a set of road sections for ensuring the driving safety of an AV-based mobility service. The main objective of this research is to analyze the safety performance of AVs on the road geometrical features present during this type of mobility service. To achieve the research goal, a mobility service is classified by a combination of six road types, including expressway, bus rapid transit (BRT) lane, principal arterial road, minor arterial road, collector road, and local road. With any given road type, a field test dataset is collected and analyzed to assess the safety performance of the AV-based mobility service with respect to road geometry. Furthermore, the safety performances of each road section are explored by using a historical dataset for human-driven vehicle-involved accident cases. The result reveals that most of the dangerous occurrences in both AV and human-driven vehicles show similar patterns. However, contrasting results are also observed in crest vertical curve sections, where the AV shows a lower risk of dangerous events than that of a human-driven vehicle. The findings can be used as primary data for optimizing the physical and digital infrastructure needed to implement efficient and safe AV-based mobility services in the future.

Recently, as a variety of position sensors are developed, a large amount of urban position data is collected in the urban traffic networks. Based on the data collected through such location sensors, high-resolution urban mobility data of individual users using urban road networks is generated and collected in the transportation systems. Urban mobility data generated by these sensors provide a novel spatio-temporal insights into the mobility patterns of traffic network users and can be used to develop models and strategies to predict traffic flows in urban areas and improve traffic efficiency. This study proposes an algorithm for predicting urban mobility patterns. Deep learning based algorithm is used to train mobility patterns in urban areas and predict mobility. The proposed algorithm is trained and tested using Bluetooth data collected in Brisbane for one year. As a result of evaluating the performance of the algorithm with the test dataset, the proposed algorithm shows an average prediction accuracy of 70% or more.