Search Content

When looking at measures of fatal and severe-injury crashes; roundabouts have demonstrated improved safety performance compared to traditional signalized intersections. Despite this; when it comes to less severe crashes; multilane roundabouts fail to provide a similar benefit. Previous research into this topic has identified behaviors that are associated with crashes in two-lane roundabouts; with yielding violations and turn violations generally being the largest contributors to crashes. This study sought to build on that work by expanding the data collection effort to include more sites and relate the frequency of individual behaviors to specific design features. Ultimately; four roundabouts were chosen for data collection and analysis; with two of them being full 2x2 roundabouts and the other two half-2x2. These locations were University Dr. S and 5th Ave. S in St. Cloud (half-2x2); 185th St. W and Kenwood Trail in Lakeville (full-2x2); and TH-22 and Adams St. (half-2x2) and TH-22 and Madison Ave. in Mankato (full-2x2). In the last two; changes in traffic control were implemented to reduce failure-to-yield crashes; and the study compared the driver behavior before and after the interventions. In general; the collective results show that the problems observed in the earlier site are present in all of the other sites with scale variations prompted by geometric and traffic control design elements. For example; in the St. Cloud roundabout; an increased rate of right-from-inner-lane turn violations were observed; which can be attributed to the sharper deflection angles present. Additionally; from the aforementioned roundabout as well as the one in Lakeville; it was concluded that turn violation rates are affected on the single or multilane geometry of the links approaching the roundabout. Single-lane roads result in fewer left-from-outer-lane violations. Overhead lane designation signs result in reduced turn violations similar to the earlier studies' interventions in the approach lane markings. Unfortunately; apart from confirming the trend; no successful design or intervention was discovered regarding yield violations.

This study evaluates recent technology that uses inductive loop detectors, traditionally used for collecting traffic volume and speed data, to provide vehicle classification data by examining the high-resolution signature produced when a vehicle passes over the sensor. The project aims to verify the accuracy of the new classification system, collect additional heavy vehicle data to help improve system accuracy, and familiarize MnDOT staff with the technology through training and the development of a field deployment manual. Through collaboration with MnDOT and the technology vendor CLR Analytics, Inc., the VSign vehicle classification system has been installed at five sites in Minnesota with preexisting loop detection systems. The final sites are representative of MnDOT facilities, feature a mix of heavy vehicle traffic, and provide accessibility for deployment staff. Data from the VSign system was compared with manually verified ground-truth data collected from video under both the Federal Highway Administration (FHWA) and Highway Performance Monitoring System (HPMS) classification schemes. The system demonstrated high accuracy for passenger vehicles but varying accuracy for different classes of heavy vehicles, though performance improved under the HPMS classification scheme. The VSign system was also evaluated against the video-based iTHEIA™ system at one site, which VSign outperformed in both classification accuracy and detection rate. The results suggested that the VSign system was more effective at locations where vehicles maintained consistent speeds and were centered in the lane due to the negative effects of variations in speed and lateral position on the consistency of vehicle signatures read by the detector.