IRI, RoadBotics and Data-Driven Pavement Management Results
By Ben Schmidt, President, Roadbotics
As the President of RoadBotics, I’ve spoken to hundreds of public works directors from communities all over the United States. I’ve also worked closely with the heads of the largest engineering firms in the world. Regardless of budget or size of road network, they all want the same things from pavement management systems; clear and accurate data to help them make informed decisions about pavement.
It’s really that simple. Yet, rapid technology advances that have made this easy and affordable within the last few years are being neglected compared with the traditional solutions that are neither simple nor accurate.
Take IRI for example. IRI was the gold-standard for pavement assessment throughout the world. Developed by the World Bank in the mid-1980s, the International Roughness Index (IRI) uses accelerometers to measure the road’s roughness along a vehicle’s wheel path. This roughness is determined by the amount of vertical movement registered by an accelerometer. A report is then provided to the road manager indicating the average roughness for every 10-100 meters depending on the Information Quality Level (IQL) being used. Based on the roughness data, the road manager is expected to make critical decisions with respect to paving.
But what information does he truly have at his disposal and how helpful will it be when determining how best to allocate his finite funds? When compared with laborious, costly, and subjective visual inspection, IRI is a marked improvement for quantitative data. However, since IRI is only capable of measuring the road’s roughness within the wheel path, there is no way it can provide a full assessment of the roadway like that visual inspection. Secondly, if IRI is only providing the average amount of vertical movement every 10-100 meters, then a lot of granularity is lost by not providing a full assessment of the roadway.
This is a tough trade-off for any road manager or authority to make when seeking to obtain information on the status of their assets. Quantitative but less granular IRI, or subjective and ideally more comprehensive visual inspections.
Similarly, since IRI does not discern what type of distress is actually on the road surface it presents challenges for taking action. Does a small amount of movement mean a transverse crack needs crack sealed? Does a large amount of movement mean a pothole that should be patched? Without knowing for certain, how is any road manager to make an informed decision on the type of treatment required for his road network. This could mean the difference between spending millions on resurfacing or thousands on preventative maintenance.
Road managers need data that is actionable and easy to understand. They need tools that can give them a rating that makes sense and the information to support that decision without immediately relying upon visual inspections.
The mission at RoadBotics has been to provide this type of road understanding in a form factor that is affordable, actionable, and safe. At RoadBotics we have created a tool that achieves this and our customers all around the world see the value in our data and can make better decisions.
So why and how is RoadBotics’ approach to road assessments better than IRI?
For starters, data collection is simple. Have a smartphone? Then you have a powerful infrastructure monitoring tool. Our customers simply use their smartphone, download our data collection app, and begin. There have been attempts to use smartphone accelerometers to calculate IRI, but there is an even better sensor in that smartphone. This is where the technology has truly revolutionized what’s possible.
We use the camera and so we see everything the visual inspection would see. Using a smartphone camera for data collection requires no vehicle calibration because you are only mounting to the windshield and ensuring that the camera is facing forward towards the pavement. Simple.
Additionally, RoadBotics rates the entire road surface because we can see it, not just the wheel path. The visual nature of the data collected by smartphone camera enables a more detailed and holistic snapshot of the pavement condition to be made – not just the wheelbase where the accelerometer data would be triggered. What’s more, those same images of the road surface can be reviewed through a comprehensive video log of the survey. This allows the road manager to make plans from his computer before going into the field.
Each image is assessed using machine learning. We have trained a computer to understand road imagery and road distresses painstakingly teaching it how to understand cracks, potholes, utility patching, etc. and to understand it at pixel-level granularity. RoadBotics trained its model by utilizing trained personnel to identify different distresses such as unsealed cracks, alligator cracks, potholes, cold patches, sealants and so on until every object in the image was categorized the way we want the computer to later categorize new images. In addition, each image was assigned a rating by a road expert. With this system, the trained model (which has been taught using millions of road images and distresses) is able to analyze the pavement distresses as well as provide a quantitative rating of that pavement.
With the rapid development of new technologies like RoadBotics, officials around the world are quickly adopting more sophisticated and data-driven approaches to asset management that were previously not possible. One of our earliest customers has seen a 10% decrease in its number of failing roads since its use of RoadBotics three years ago. The simplicity of data collection coupled with the reasonable cost associated with a RoadBotics assessment, allows for more frequent assessments and comparative annual data for road managers to make informed decisions on road maintenance.
RoadBotics uses artificial intelligence and smartphones to help governments and engineering firms make data-driven decisions. For more information, please visit www.roadbotics.com[/column_1] [/column]