Abstract
Intersections are dangerous. In the U.S., approximately 40% of all crashes, 50% of serious collisions, and 20% of fatalities occur in intersections. Intersections are challenging due to complex interactions among pedestrians, bicycles and vehicles; absence of lane markings; difficulty in determining who has the right of way; blind spots; and illegal movements (e.g., vehicles running red lights). Many cities have developed Vision Zero plans seeking to eliminate traffic injuries and deaths through modifications to road infrastructure, such as adding bike lanes and pedestrian refuge islands. These modifications can be expensive (e.g., the cost of a protected intersection can range between $250,000 to more than a $1 million) and have mixed safety results.It is claimed autonomous vehicles (AVs) will prevent 94% of all crashes involving human error. However, the safety performance of AVs is far below that of human-driven cars. In California, the number of accidents and disengagements per AV mile traveled is 13 to 100 times worse than human-driven cars. The AV fatality rate is equally as bad. AVs find intersections especially challenging; 58 of 66 (88%) AV crashes reported to the California Department of Motor Vehicles (DMV) occurred in intersections.Crashes in intersections occur because vehicles, pedestrians, and bicyclists are missing critical information. Intelligent intersections can provide this information at a relatively low cost of $25,000 to $100,000 per intersection. Intelligent intersections are able to report the traffic signal from all approaches, predict when the signal phase will change, relay information on blind spots, predict red light violations before they occur, and more. This information is broadcast via radio to every traveler in the intersection equipped with a smartphone or Bluetooth device.
