Developing Alternative Technologies to Detect Traffic at Intersections
What makes the traffic light turn green at some intersections? You may know that there are sometimes sensors installed under the pavement that can detect traditional motor vehicles and trigger the light to change. These systems aren’t perfect, though. For example, they are expensive to install after the pavement has been laid. Technological advances are making nonintrusive systems (NIS) possible as an alternative and transportation departments, including the Minnesota Department of Transportation’s Metro District, are taking a good look.
Yao-Yi Chiang, an associate professor in Computer Science & Engineering and a CSE DSI affiliate, is now leading a study initiated by John Hourdos (previously a Research Associate Professor at the University of Minnesota and currently a Highway Research Engineer with the Federal Highway Administration) and sponsored by MnDOT that is identifying which attributes of nonintrusive systems and in-ground sensors local governments should consider when deciding which traffic detection system is right for their conditions. These aren’t clearcut decisions since every intersection has its own unique location context (e.g., traffic and weather patterns).
Traditional underground loop detectors have been aimed at detecting traditional motor vehicles - i.e., large steel objects. They consist of a continuous length of wire that resonates at a regular frequency. When a car or truck — something big and metal — drives over it, the frequency increases and the circuit that is usually open now closes. A computer controller uses this signal to decide when to trigger the light to change. However, these systems aren’t as good at detecting bicycles, motorcycles, and EVs.
New nonintrusive systems include video- and radar-based detection systems that are installed above ground. Under Hourdos’ direction, the researchers on the project began by examining nonintrusive systems that are currently available and researching the literature on them. Then they interviewed traffic operations staff at Minnesota transportation agencies that are already using NIS systems.
According to Chiang, “The interviews by Dr. Hourdos provided valuable insight into the real-world experiences of Minnesota transportation agencies, including the benefits, challenges, and costs of operating and maintaining the different systems.”
After comparing the NIS with in-ground sensors by looking at ten Twin Cities intersections under different environmental conditions, the team found that various types of weather and environmental factors can have a big and significantly different impact on the performance of nonintrusive systems at different locations. For example, glare, snow, ice, and dirt can accumulate on the camera lens and interfere with the effectiveness of the system.
The study found that no system ranked first in all conditions, but some were better in certain conditions. NIS performance can be improved by installing shields or filters to reduce glare and heat shields to protect the devices, repositioning the cameras when necessary, and using a central monitoring system so that checks can be performed remotely to diagnose any problems. In addition, NIS installation costs were significantly less than the cost for traditional loop detectors. This is because if the road is already finished when the in-ground sensors need to be installed, the existing pavement has to be cut through and then replaced.
The research team used their findings to develop a new decision tool that includes both traditional in-ground sensors and nonintrusive systems. This tool is based on huge amounts of data from various sources, including traffic cameras, traffic controllers, weather stations, and traditional census and survey data, to help operators select, install, and maintain the NIS product that suits their needs and conditions and reduce maintenance costs.