We were able to collect 400-500 miles worth of Ground Truthed data with a single vehicle over 10 days. (That's 75% of Syracuse's entire street grid)
For Pittsburgh, we made a conscious decision to focus on bike lanes because:
1. It moves the needle farther for a specific user group (bike commuters). 2. Better suited to win hearts and minds of traditional public works managers who can be skeptical of this approach. 3. Bike lanes are a throwback to SQUID's origins when Patrick and I rode 100 miles in NYC collecting bumpiness data in the fall of 2014. More about the SQUID story here: https://www.hackster.io/argo/squid-street-quality-identification-device-a43367 4. More suited to a community engagement angle where you can "adopt your bike lane" by using Open Street Cam to survey the bike lane outside your house.
re: efficient road repairs: The conventional response paradigm uses on a combination of PCI (Pavement Condition Index) scores, user complaints, and perhaps other unknown criteria. The PCI standard was developed in the 1970s and relies on eyeballing road conditions and assigning an overall score. This is far too subjective and prone to error. Furthermore, pothole fixing crews are routed in a whack-a-mole manner which is not necessarily the most efficient allocation of maintenance resources.
I'd rather use the term anticipatory vs the rather loaded "predictive maintenance". This is something we are eager to prove out. The marginal cost of collecting high quality data frequently about all streets in a city has reduced exponentially and this can get cities out of a "whack-a-mole" response paradigm and into a more proactive one which we believe will lead to more efficient (and data driven) allocation of maintenance crews.