Project

Purpose and goal: The construction of oncoming separated rural roads with intermittent overtaking sections, so called 2+1 roads, in lieu of traditional two-lane undivided roads, has the potential to improve safety by minimizing the risk of head-on collisions. It is considered as an important part in achieving Vision zero by 2030 and the ambition is therefore to build 150-200 km of 2+1 roads every year. In several respects, overtaking behavior is unique on 2+1 roads as there are predefined sections where overtaking can take place. Therefore, with the increase in reconstruction of two-lane undivided roads to 2+1 roads, there are a growing interest in questions related to the design of such roads, including the length and placement of overtaking lanes to improve both safety and traffic efficiency.

Traditionally microscopic traffic simulation has been used as a tool to investigate the impact of different designs of 2+1 roads. However, the commercially available traffic simulation models are mainly designed to simulate urban roads and motorways, and in previous studies the calibration of the microscopic traffic simulation models for 2+1 roads have only been case-specific and based on aggregate metrics. There is limited knowledge to confirm if the driver behavior models successfully replicate the same on 2+1 roads.

A recent research project conducted with funding from the Swedish Transport Administration applied a novel vision-based methodology to gather evidence about vehicle speeds and overtaking and merging behavior along a 2-lane segment on the E18 route. The new dataset also created the opportunity to assess the capability of different microscopic traffic simulators (SUMO and RuTSim) in replicating the driving behavior. Beyond the scope of the existing project, there is a further need to disseminate the novel vision-based vehicle recognition method and the resulting dataset. It is also important to present the most interesting findings from the analysis and explore future developments of the driving behavior models in microscopic traffic simulators. There is also a need to extend the investigation as the present study has only been conducted on one 2-lane section of a 2+1 road as a case study.

Planned approach and activities: Based on the existing analysis and some additional data analysis, one (or two) manuscript(s) shall be prepared. The focus of the first manuscript will be on a new method adopted for data collection and vehicle recognition and subsequent data analysis. The second manuscript will be focused on applying the RuTSim simulator (developed in VTI) and comparing the outputs with the real-world data as well as with a commercially available and widely used microscopic traffic simulator, SUMO. The second manuscript may also require some additional simulation runs above and beyond the scope of the ongoing research project. The target journals shall be: Transportation Research Part C: Emerging Technologies, European Transport Research Review, Accident Analysis and Prevention, Transportmetrica B: Transport Dynamics, and/or Transportmetrica A: Transport Science.

Another activity will be to brainstorm on the future scope of work from the currently gained new knowledge. There will specifically be two directions: (1) how to apply the recognition method on other 2+1 roads to study the change in road user behavior with change in road configuration. (2) How to develop, modify and/or calibrate existing driving behavior models to replicate the unique driving behavior along 2+1 roads.

Expected results: The expected results will be to consolidate the important findings from the recently ending project and submit it for possible publication in one (or two) high-quality international journal(s).

A plan for future projects based on extensions and augmentations of the current work will also be developed as a part of this idea exploration.

Contribution to SAFER: The completed data analysis brings out to a great extent the unique nature of overtaking behavior on 2+1 roads that includes evidence about speeding and aggressive merging near the end of the 2-lane segment. These findings are interesting both from safety and traffic efficiency perspectives and the proposed study will give new directions for developing and calibrating simulation models for investigating different configurations of 2+1 roads. The added knowledge can be important to achieve the target of Vision Zero.

The vehicle recognition method and the resulting novel dataset and its potentials will be interesting for performing more advanced and detailed studies on road user behavior along a longer length of roads than is currently feasible.