An Evaluation of Vehicle Identification Technology

To successfully manage the state road network, MnDOT needs a thorough understanding of the number and type of vehicles on the road. To obtain this information, the agency upgraded existing inductive loop infrastructure at select locations to enable these sensors to collect vehicle classification data. This project evaluated the accuracy of the inductive loop upgrade and its life cycle costs to determine its viability for future use on Minnesota roads.

The number and type of vehicles traveling on Minnesota roads significantly impact MnDOT management decisions about transportation system planning, forecasting and funding. To obtain vehicle information, MnDOT uses tools such as weigh-in-motion sensors and automatic traffic recorders, as well as manual data collection practices. Inductive loop infrastructure has been installed at over 300 locations in the state. MnDOT needed a cost-effective method for collecting this information from additional locations.

Upgrading the hardware and software of existing infrastructure could provide a cost-effective option for collecting vehicle classification information. The upgraded inductive loops include sensors that receive feedback from metallic materials on the undercarriage of traversing vehicles to create a high-resolution signature of vehicle types. For example, passenger cars have thinner axles and engines that are lower to the ground. Large trucks have thicker axles and engines that are higher above the ground. The software uses these characteristics to create signatures that classify the vehicles.

Table of Contents

What Did We Do?

This project aimed to evaluate the performance of inductive loop sensors in observing and classifying vehicle types, and to develop installation procedures for future implementations of this technology. To complete these objectives, investigators:

Installed new detector cards, processing hardware and cellular modems at five sites with existing inductive loops. The sites represented different deployment scenarios and infrastructure types.
Collected data from these sites, which classified vehicles according to the Federal Highway Administration’s (FHWA’s) 13 vehicle classes.
Compared the inductive loop data and classifications to manually collected data and classifications.
Compared the inductive loop classification performance to the performance of another commercial vehicle classification system.
Performed a life cycle cost analysis to compare the cost of the inductive loop system to the conventional vehicle classification systems that MnDOT typically uses.

What Was the Result?

The inductive loop upgrades demonstrated a relatively high level of performance, classifying vehicles with an average accuracy of 95%. Performance was highest for passenger vehicles, with an accuracy of 99% for Class 2 vehicles (passenger cars) and an accuracy of 91% for Class 3 vehicles (slightly larger vehicles such as ambulances and light-duty trucks).

“These research results have validated an additional option for MnDOT to collect vehicle type data from significantly more locations,” said Gene Hicks, director, MnDOT Traffic Forecasting and Analysis.

Performance was lower when classifying trucks, with an average accuracy of 63%, 70% and 87% for Classes 5, 6 and 9, respectively. In these cases, the technology had difficulty differentiating between similar classes of trucks. For example, when the inductive loop misclassified single trailer trucks, the technology primarily misclassified them as a similar truck class. Reducing the number of vehicle classes from FHWA’s 13 classes to the Highway Performance Monitoring System’s seven classes resulted in a 97% accuracy in correctly classifying single trailer trucks.

When compared to the performance of another commercial vehicle classification system, the inductive loop technology performed better in vehicle classification (92% vs. 86%) and detection rate (100% vs. 77%). Overall, the inductive loop technology upgrade performs best when vehicles are moving at a consistent speed and are centered in the lane.

A preliminary life cycle cost analysis indicated common scenarios in which the inductive loop technology would be cost-effective despite ongoing software subscription and hardware maintenance costs. For example, if funds from a different program maintain an existing inductive loop, this may shift the cost analysis in its favor.

What’s Next?

To improve accuracy, MnDOT is working with the inductive loop vendor to assess potential adjustments to the classification processes for a subset of vehicle signatures. Additionally, federal testing of this technology is underway in other states and may provide valuable insight to MnDOT. While there is no current plan to implement beyond the five test sites, expansion is possible based on how costs shift over time.