1. Home
2. Ideas
3. Traffic Sensor Testing Methodology and E...

Traffic Sensor Testing Methodology and Evaluation Program

• Jul 29, 2020

Research Description

There is rapid growth in new ITS sensor system products that leverage new technology(s). We are seeing new types of sensors focused on machine-learning, artificial intelligence, and automation of results, which are lacking in any independent, ground truthing to verify results and quantify error. This creates a tremendous agency burden to test each sensor type and iteration that comes to market, often with disappointing results.

Manufacturers (and their distributors) of ITS sensor products rarely, if ever, can provide independent 3rd party field verification, validation testing results to demonstrate the “real-world” effectiveness of their sensor(s). Sensors are increasingly being utilized in the operation of traffic and ITS roadside devices of which many function in highly sensitive contexts that could potentially result in device operational error and/or failure.

The industry typically relies on a “try before you buy” business model of offering agencies low/no cost access to products. This approach relies on valuable, limited agency resources to field test to assess performance and usefulness. Often our agency tests are not rigorous and limited in their scientific approach to gauging sensor(s) performance. Thus, the transferability of one agencies experience is limited, and a significant amount of agency resources are expended nationally in this “try before you buy” model. There is a need for a new model that is useful, transferable, and resource efficient. This research would establish a consistent ITS sensor evaluation methodology and test protocol(s), such as error types, thresholds and basis for evaluation. The methodology will need to consider environmental conditions. Test protocols will need to be developed for each mode being detected (e.g. vehicle, pedestrian, bicycle). This will allow national research laboratories to evaluate a sensor(s) with confidence that the methods and protocols are replicable, transferable, and useful to transportation public agencies and their partners. This will save agencies a tremendous amount of costs and resources if they can rely on the sensor testing done elsewhere. AASHTO’s National Transportation Product Evaluation Program (NTPEP) has indicated an interest in expanding into this type of ITS sensor evaluation if a test protocol was available. The research team would need to develop an implementation guide for the sensor evaluation to be implemented within the NTPEP (or similar) program.

Additional Supporting Information

In the process of developing this research idea, the following agencies/organizations have expressed interest & support the need for this work: • Oregon Department of Transportation • Georgia Department of Transportation • Utah Department of Transportation • Indiana Department of Transportation • Purdue University • Oregon State University • Washington County, Oregon • Clackamas County, Oregon • Portland Bureau of Transportation

---

Submitted By

Galen McGill

Oregon Department of Transportation

503-986-4486

Attachments