< Back to Session View

A Simulation-Based Framework for Assessing Operational and Safety Benefits of CAVs

Date and Time: Tuesday, July 11, 2023: 5:30 PM - 7:00 PM

Bryce Grame 

PhD Student, University of Florida

Presentation Description

The University of Florida Transportation Institute (UFTI) in collaboration with the Florida Department of Transportation (FDOT) and the City of Gainesville (COG) has developed the I-STREET initiative to test emerging connected and autonomous vehicle (CAV) technologies with the aim of improving safety and mobility of travelers. I-STREET is an instrumented network of streets in Gainesville, Florida, used as a “living lab” to research emerging transportation technologies.
A key I-STREET aspect is the “Gainesville SPaT Trapezium corridor” in which 27 signalized intersections are deployed with roadside units (RSUs) and other controller upgrades to deploy and test connected vehicle (CV) technologies. RSUs are deployed at all signalized intersections for vehicle to infrastructure (V2I) communications using dedicated short range radio communications (DSRC). These can be used for detection and broadcasting of standardized vehicle to everything (V2X) messages like Basic Safety Messages (BSMs).
The Trapezium corridors represent the roadways providing access to the UF campus, which are heavily congested during the peak periods. There have been over 10,000 traffic crashes along these corridors over the last 10 years, and many of these are at or near intersections. Therefore, improving traffic flow and reducing crashes along these key arterials will immensely benefit the mobility and safety of Gainesville residents. The advent of CAVs and the instrumentation (RSUs) added to the infrastructure provides opportunities to (1) collect new data like near misses at intersections and to (2) explore the impacts of new countermeasures such as provision of CAV safety messages.
This research in progress provides a systematic procedure for (1) determining new performance measures for safety using vehicle trajectory data and (2) evaluating the impacts of information-based countermeasures. The changes to traffic operations metrics like queue lengths and delay and surrogate safety measures like near misses at intersections are examined. Further, the analysis examines impacts on a system of intersections along two major arterial road systems.
All analysis is performed in a simulated before-and-after study environment, as the current levels of market penetration of CAVs remain too low to support field experiments. The simulated vehicle fleet composition is determined by the vehicle types and the percentage of each type in the fleet. Four types of vehicles are considered in this study: (1) normal human driven vehicles (NV), (2) human-driven connected vehicles (CV), (3) conservative connected-automated vehicles (C-CAV), and (4) aggressive connected-automated vehicles (A-CAV). All connected vehicles (types 2-4) were subject to speed control strategies. Both conservative and aggressive CAV’s (types 3 and 4) had altered driving behaviors like adjustments to gap acceptance and car-following models. Both local and corridor-wide speed controls were implemented to replicate DSRC and cellular-vehicle-to-everything (CV2X) communication scenarios, respectively.
Two corridors with very distinct characteristics were modeled, representing the east and west sides of the trapezoid that forms the Trapezium. While both corridors have 7 signalized intersections, the east corridor is only one mile long and has a speed limit of 30 miles per hour (mph), where the west corridor is around two miles long and has a speed limit of 45 mph. This gives the east and west corridors signal spacing densities of approximately 7 signals per mile and 3.5 signals per mile, respectively.
10 simulations were taken for each experimental run, simulating the peak hour of traffic conditions on the corridors. This allowed for average values of operational and safety metrics to be compared across experimental runs to average values from the calibrated base models of the corridors. Including the two calibrated base models, a total of 55 of these sets of 10 simulations were run. The remaining 53 experiments varied the percentage of NVs, CVs, and CAVs in the fleet, as well as the three speed control strategies (no speed control, local speed control, and corridor-wide speed control). To replicate realistic ranges of DSRC communications, only a corridor-wide speed control strategy was implemented on the east corridor.
This effort is relevant to road vehicle automation because simulation findings show significant implications for traffic operations and safety in a mixed fleet environment. Overall, results indicate that reductions in speed did not cause significant increases in delay for any of the intersections. While it could be hypothesized that slowing vehicles down should increase delay, it should also be noted that at slower speeds, vehicles may be in a “queuing state” for less time (in contrast to traveling at a higher speed and quickly reaching the end of the queue). As delay is measured over the time spent by the vehicle in a queue, this can also affect the delay calculations.
The impacts of speed control are more apparent when examining the corridors’ travel times. In general, the average travel times are shortest with no speed controls and longest with corridor-wide speed controls. The travel times for aggressive CAVs are less than the travel times with conservative CAVs, as would be expected. Nonetheless, the magnitude of the travel time increases do not appear to be significantly large.
In general, the number of total conflicts and rear end conflicts decreases with an increase in the percentage of CAVs in the traffic stream. This trend is also seen in the no speed control scenario. In fact, the decrease in number of conflicts for the no-speed control scenario relative to the base case is generally higher than the decrease in conflicts in speed control scenarios relative to no-speed control scenarios. This suggests that the safety benefits are more because of automation (homogeneity and instantaneous response times) rather than connectivity and safety messages (speed control in this case).
On examining vehicles by type, conflicts involving two NVs are the most frequent. Among conflicts involving CAVs and NV, there are more cases in which the NV is behind the CAV. This is reasonable, as the NV are unable to quickly react to the maneuvers of the CAVs. Conflicts involving two CAVs are the least frequent. Similar to other operational studies on CAVs, this research suggests that until a significant fleet percentage of CAVs can be achieved, there may be negative impacts to both operations and safety.

Speaker Biography

Bryce holds a Bachelor's of Science in Civil Engineering and a non-teaching statistics minor from Montana State University, where he graduated with Summa Cum Laude distinction in 2021. He is currently enrolled in a concurrent-degree program at the University of Florida, which allows him to count credits towards both a Master of Arts in Urban and Regional Planning and a PhD in Transportation Engineering. Bryce is interested in the potential that new technologies like connected and autonomous vehicles hold to improve safety in transportation. His research bridges planning and engineering, addressing safety from an interdisciplinary perspective. When he finishes graduate school at the University of Florida, Bryce would like to work as a hybrid 'plangineer' addressing safety concerns in transportation across the planning and engineering professions. After passing the PE exam in the Fall of 2021, Bryce would like to now gain AICP certification after he graduates from his planning program in the Spring of 2024.

Presentation File