Stern, Raphael

Max-pressure (MP) traffic signal control is a new and innovative control algorithm that uses upstream and downstream vehicle counts to determine signal timing that maximizes throughput. While this method has been extensively tested in simulation, it has not yet been tested on actual traffic signals in the US. To close this gap, this report presents the results of the development of a hardware-in-the-loop traffic signal testbed where microsimulation is used to simulate realistic traffic conditions, and the MP algorithm is used to control the signal display using a traffic controller (Q-Free MaxTime controller). The hardware-in-the-loop results demonstrate that MP can be safely deployed on North American traffic signal control hardware.

People who walk and bike are the most vulnerable road users. However, understanding where they walk and bike requires continual data monitoring. Traditional methods rely on physical sensors in the infrastructure to detect the presence of pedestrians and bicyclists. However, these are expensive to deploy and only detect road users at the specific locations they are deployed. Instead, this study develops methods to use mobile phone based GPS data to estimate the number of bicyclists and pedestrians, and applies this methodology to the Twin Cities Metro area in Minnesota. The developed methodology is able to estimate average pedestrian and bicyclist volumes with relatively high accuracy.

Many factors influence an individual driver’s decision to yield or not yield to individual pedestrians attempting to cross the road at an unsignalized crossing. This study collects observational data from more than 3,300 crossing events at 18 intersections in Minnesota to further our understanding of what factors positively influence driver yielding. Using the collected data, a statistical analysis was conducted to identify features that most strongly correlate with driver yielding. Event specific features such as speed were found to greatly influence yielding, with vehicles traveling at a speed of greater than 25 mph significantly less likely to yield to pedestrians than vehicles traveling at speeds lower than 25 mph. Site-specific features such as the presence of signs indicating a crossing were also strongly correlated with driver yielding. The results provide indication of which features of unsignalized crossings correlate with higher driver yielding rates. These findings can be used to guide policy and design at sites where a high driver yielding rate is desirable.

In 2009, the FHWA’s Manual on Uniform Traffic Control Devices (MUTCD) introduced the flashing yellow arrow (FYA) traffic signal as an alternative to circular green (CG) to indicate permitted left turns. The FYA is arguably a more intuitive indication that left turns are permitted but not protected and, in addition, the FYA signal heads can support time-of-day changes between protective and permissive left -turn phasing. In 2019, a Research Needs Statement stated that “Research is needed to examine driver comprehension of flashing yellow arrows in different light arrangements and the role of signage.” Our objective in this project was to assess drivers’ understanding of FYA signal indications and to see if the presence or absence of “Left Turn Yield” signs affect gap acceptance. This was accomplished by conducting an online survey of drivers regarding their understanding of FYA signals and by carrying out a field study of drivers’ gap acceptance at a set of Twin Cities intersections.

Deer-vehicle collisions (DVCs) represent a significant hazard on Minnesota roads, with roughly 1,200 DVCs reported annually to the Minnesota Department of Public Safety (MnDPS) and many more going unreported. While DVCs are common across Minnesota, local variations in deer density as well as roadway characteristics and use patterns make DVCs more likely to occur on some roadways than others. Moreover, the true extent of DVC concentrations is unclear due to the high proportion of DVCs that go unreported. This report presents findings from research that (1) uses data to identify areas of DVC concentration based on the specific roadway characteristics and (2) presents a methodology to estimate DVC reporting rate across the state. This methodology is applied in a pilot study in the Duluth area, as well as in an extended search area that includes highways spread across much of outstate Minnesota to estimate the DVC reporting rate.