What do the ratings mean?

What do the ratings mean?
January 17, 2019 Ryan Gayman

This is an excellent question that we often hear from first time customers. In this blog post, we break down where the ratings come from and how to interpret them for your pavement management needs.

From images to pavement ratings

RoadBotics ratings start with high resolution images of our customers’ entire road networks. Our teams of operations technicians collect these images by driving the full extent of a customer’s roadways using standard passenger vehicles equipped with windshield-mounted smartphones. The smartphones capture images of the pavement surface via our proprietary RoadSense data collection app.

We take these images from the field and extract high quality images for every 3-meters (10-feet) along all the roadways in a customer’s network. These images are then transferred to our machine learning process where they are assessed pixel-by-pixel and automatically assigned a 1-5 condition rating score. Each rated image, called a “Point” is then GPS snapped to the interactive RoadWay platform map. In this way, customers can login to RoadWay and view the rated images of their entire road  network.

So what do the RoadBotics ratings mean?

Here are the basics to know about the RoadBotics ratings:

Level 1: No or very minor surface damage. This is likely a road that was paved with in the past 3 years. This is a text block. Click the edit button to change this text.

Level 2: Minor surface damage, no critical issues. This road is beginning to show typical wear and tear.

Level 3: Appearance of pervasive surface distresses. Important issues in the surface are beginning to show.

Level 4: Significant damage or emerging critical failures. The road damage is beginning to exacerbate.

Level 5: Major surface damage and/or critical surface failures are present on the road surface.

RoadBotics machine learning process automatically generates these 1-5 ratings for each image data point it analyzes based on these characteristics:

  • Type of distresses present
  • Frequency of appearance of distresses
  • Density of distress composition in the image

The numeric ratings are often described as integers (1, 2, 3, 4 or 5), but the actual distribution of their value is defined to the tenth decimal. The highest bound of the rating system is 1.00 and the lowest bound is 5.00.

Level 1 “Green” = 1.00 to 1.49

Level 2 “Light Green” = 1.5 to 2.49

Level 3 “Yellow” = 2.5 to 3.49

Level 4 “Orange” = 3.5 to 4.49

Level 5 “Red” = 4.5 to 5.00

This rating system is based on PASER – the standard method and 1-10 scoring model used by trained civil engineers for visually inspecting the condition of road surfaces. PASER ratings are generated by a civil engineering walking or driving along a roadway, viewing the pavement surface and making 1-10 judgement about the condition of the road surface based on a standardized, pre-defined criteria. Unlike PASER, where the rating reflects the quality / condition of the road (1 is a poor road), the RoadBotics scale reflects the level of distress (1 is a high quality, newer road surface).


Reading Points vs. Segment data

Point Data – Remember that point data represents images that were collected and initially rated 1-5 via RoadBotics’ machine learning process. Points represent approximately a 3-meter (10-foot) path of roadway. Points are snapped to their location on the RoadWay mapping platform based on their GPS coordinates. When you click on a point in RoadWay, you see the image that was captured, the color-coded and numeric rating generated for that image, timestamp of when the image was taken and the GPS latitude/longitude.