Rural Intersection Safety for Autonomous and Connected Vehicles (USDOT)

Project summary:

About 75% of all roads in the United States, around 3 million miles, are in rural areas and are vital for transporting goods and connecting communities. The likelihood that a car crash will result in death is 62% higher in rural America, even with less than one fifth of the population living in these areas. Approximately 85,000 people were killed on rural roads between 2016 and 2020. Locally, in the state of Minnesota, nearly two-thirds of all crashes leading to fatalities or serious injuries occur at rural intersections. This project will develop and demonstrate novel technology that can help autonomous vehicles (and other connected vehicles in general) safely handle unsignalized rural intersections. The vehicles will both be able to travel through the intersection and make safe turns at the intersection as desired. The typical sensors on an autonomous vehicle do not have the field-of-view to be able to monitor the trajectories of distant vehicles in cross-traffic and correctly identify vehicle gaps for safely traveling at thru-STOP rural intersections. An intersection-installed unit consisting of a vehicle sensing system and wireless hardware that will enable an intersection to communicate with connected and autonomous vehicles (CAVs) will be developed. The key components of the proposed system will include a low-cost (~$170) radar sensor chip and antenna, microprocessor-based software for accurate trajectory tracking of vehicles and an infrastructure-to-vehicle wireless system for communicating with CAVs. The proposed system will help CAVs handle rural intersections safely and efficiently, including those with missing delineation and signage. It can also be integrated into recent intelligent intersection warning systems implemented in Minnesota that provide visual alerts (appropriate flashing lights) to stop/allow safe passage of human driven vehicles at thru-STOP intersections on today?s rural roads. The proposed system is low-cost and suitable for widespread deployment. The novelty of the proposed technology lies in the algorithms that have been developed to correctly cluster and classify radar reflections so as to recognize moving vehicles accurately and to track the lateral and longitudinal positions, velocities and orientations of vehicles using intelligent algorithms. The capabilities of our algorithms enable the use of an ultra-low-cost radar device designed for use in static installations (e.g. at a stop sign) that returns raw reflection measurements from objects. Tasks in this one-year project include developing comprehensive vehicle tracking algorithms for use at an intersection using a low-cost radar chip, implementing a roadside hardware unit containing all sensors, electronics, and wireless communication hardware, interfacing with the autonomous MnCAV vehicle, and conducting a public demonstration of the technology using the MnCAV vehicle at a rural intersection where the system is installed.