Our Recent Posts



No tags yet.

The sad tale of Pierre the Pothole

In post-WWII Japan, it was an American, W. Edwards Deming, who championed the simple but powerful notion that if you can’t measure it you can’t manage it and Japan’s economic recovery was born. For Deming, variations and defects were not inherently bad. On the contrary, while they were to be avoided they were also useful clues when they happened and it is often the smallest anomalies – if and when you can see them – that point you directly to the largest sources of your problems and can also tell you what to do about it.

Thanks to Deming and his ideas, our cars can easily last 5 or ten years, companies can produce products at a rate of one defect per +3 million and planes can fly safely for countless miles. Wouldn’t it be great if we could readily use and easily apply Deming’s principles of continuous monitoring to our world’s roadways?

Well, a Carnegie Mellon University spinout, called RoadBotics, has made that possible. I’ve often wondered if I would someday live in a world free of potholes. I think that is now within reach. RoadBotics has married a standard smartphone app, called RoadSense, with a powerful AI-driven, cloud-based analytic platform, called RoadWay, that sees, characterizes, and assesses an ever-growing array of road surface defects, anomalies and problems as well as other roadway features (i.e., signs). All you need to do to make it work is download the app and get a set of wheels to drive and you’re good to go.

Carnegie Mellon is arguably the world’s leading computer science and robotics program and has given birth to an astonishing array of new roadway-related companies including, most recently, Uber’s autonomous vehicle lab and Ford’s recent announcement of a $1B investment in autonomous vehicle research facility, both based in Pittsburgh and the HQ of RoadBotics. So, it seems natural that from this pedigree RoadBotics would have an innovative focus on the vexing problem of effective and efficient pavement preservation.

Their basic approach is simple but the platform relies on an integration of some very sophisticated technology. They exploit the video camera, gyroscope and accelerometer capabilities of standard, run-of-the mill smartphones to capture meter-by-meter data about the road surface specifically and roadway generally. This data is captured in the phones internal memory and then uploaded to their cloud-based, AI-driven platform for immediate analysis, where key surface features like potholes, cracks, and other anomalies, along with signage are characterized and assessed.

In the cloud, the roads features are identified and 3-meter section of roadway is graded according to the damage and general wear and tear observed by using a proxy of industry standard and commonly used road surface and roadway performance standards. This information is then rendered in the form of an interactive map (see image) that provides, at a glance, a view of current road surface conditions in a color-coded rating system. The color coding can be adjusted to the pavement rating system that you are applying. Once the analysis is complete, the data and video along with the map-based visualizations are available to the client, all through the Internet.

What is particularly exciting about this development is the possibility to revolutionize data collection approaches for pavement management. Current approaches include manual ‘walking’ inspection, windshield surveys where routes the driver and a passenger-observer view the road surface and, if you afford it, a high-grade sophisticated sensor equipped vehicles. RoadBotics does not necessarily seek to replace these approaches but rather augment them. Building the RoadBotics approach into your pavement management process would enable an initial course sort of the good the bad and the ugly with respect to potholes.

This can serve a foundation for closer inspection, or could even be the launching point for rapid response to potholes that can be dealt with quickly.

It seems to me that one of the important things in improving the user experience with respect to roads and highways, is to let the public know that you’re listening and responding to their needs and to the reports that they provide you. What better way than having a rapid overview of the condition of your road network and having the ability to choose selected target potholes for immediate extermination?

This approach also lends itself to hybrid data collection. The first element could be for RoadBotics team to collect the data by conducting a video survey of your roads together with a second element of crowdsourcing data collection. Any travelers could download the app and act as a roving sensor. The term “pothole probes” comes to mind.

The third element is what I would call “fleetsourcing”. There are many fleets of vehicles traveling our highways at any given time. These include rental cars, street sweepers, US Postal Service vehicles to name a few. Using this technique, it would be possible to equip an entire fleet to act as probes while they go about their normal day-to-day business.

I expect all three elements will be combined to offer a rapid and complete assessment of an entire state within a reasonable period.

The other exciting thing about this approach is what I would call the “dynamics of deterioration”. The approach is cost-effective enough for the video surveys to be conducted three or four times a year.

Using the appropriate locational referencing technology and the artificial intelligence embedded into the RoadBotics back office, it will be possible to observe the growth and decay of roadway damage. This would enable us to consider the rate at which the pavement is deteriorating and improve our predictions to assist in better targeting resurfacing and pothole repair work.

Who knows, with the help of RoadBotics, we might become so familiar with our potholes in the future that we can give them names?

I would like to see the following obituary of my dear old friend before my own:

“Our old friend Pierre the pothole is no more. The local municipality dealt him a fatal blow with the use of a shovel and some asphalt. Thanks to improved payment monitoring techniques Pierre had a very short, yet well-documented life. We knew him well and watched him grow from an insignificant depression, until he became a hole in the ground. He was ‘nipped in the bud’ just before his prime. We witnessed his short, yet illustrious career in full detail. His demise was expected, but there is growing concern among the pothole community regarding a rapid decline in life expectancy.”