Automated Pothole Detection
We used data from Mercedes-Benz vehicles to identify needed roadway repairs
The Analytics Team and the Mayor’s Office of New Urban Mechanics conducted a pilot with Mercedes-Benz USA to use anonymized data from Mercedes-Benz vehicles to identify potential potholes. This experiment explored whether this data could usefully supplement potholes reported by residents through 311.
Names of Contributors: Betty Fang
Year(s): 2023
-
Still have questions? Contact:
Analytics Team
-
1 City Hall Square, Room 703
Boston, MA 02201
Why We Did This
311 is a valuable data source for hearing directly from residents and understanding where services are needed.
However, we know that not every issue gets reported, and are always looking for ways to collect and leverage data to give departments the most comprehensive understanding possible.
Mercedes-Benz vehicles are equipped with advanced suspension systems that register sudden compression and rebound. The Mercedes-Benz USA team processed this data and provided the City with the locations of potential potholes. We wanted to know whether this data could be used to identify potholes unlikely to be reported otherwise, or if this could help us identify potholes sooner when they might be easier to repair. Above all we also wanted to be sure the reports provided by Mercedes-Benz corresponded to an issue that our Public Works Department could fix, so that we did not divert resources into investigating false positive reports.
What We Did
We had three key questions in our analysis:
- Where are we most likely to receive reports from Mercedes-Benz? Are these reports distributed across the City? Is the overall geographic distribution similar or different from 311 reports?
- Are the Mercedes-Benz reports more often previously unknown potholes, or duplicates of potholes reported through 311?
- What proportion of the Mercedes-Benz reports are actually potholes that the Public Works Department could repair
Answering these questions involved deduplicating both datasets, and incorporating other data such as ownership, speed limit, and width of each street segment. Mercedes-Benz provided reports of potholes within the geographic boundaries of Boston but we were particularly interested in reports on roadways maintained by the City.
We also visually inspected a sample of the Mercedes-Benz reports to better understand what types of roadway conditions were likely to appear in the data and how the severity rating provided by Mercedes-Benz compared to what we saw on the ground.
Results
We found that a similar proportion of reports from both sources came from each neighborhood. However, the 311 and Mercedes-Benz data was more complementary than we expected, helping us find potholes on different types of streets. Reports from Mercedes-Benz were less likely to come from small, one-way streets. Our hypothesis is that on those roads it is easier to avoid the pothole, which means the car would not register it. The Mercedes-Benz reports were particularly helpful for identifying potential potholes on larger streets with higher speed limits, where 311 reports are relatively rare.
When we examined the reports in person, it was clear that other types of irregularities in the road were getting picked up, such as depressions near utility covers. This is still valuable information about the state of the roadways, but being able to distinguish between different causes is important as the responsibility for fixing the issue can fall with different entities.
We provided our analysis and feedback to Mercedes-Benz, helping them identify potential improvements to the technology and approach, and look forward to future opportunities to systematically gather new data on the conditions of the roadway.
You can read more about this project in the press release from Mercedes-Benz USA
Who's Involved
Who's Involved-
Still have questions? Contact:
Analytics Team
-
1 City Hall Square, Room 703
Boston, MA 02201
