Startups collecting data about Pittsburgh roadways
Groups think they can reshape infrastructure
According to the failed Amazon HQ2 bid’s description of life in Pittsburgh, “Commuting is easy.”
“Eight out of 10 Pittsburghers live within walking distance of our 53-station light rail network, 20mile busway system, and a robust network of 70+ bus lines throughout the region,” the proposal claimed.
Pittsburghers’ average commute is 26 minutes each direction — much shorter than in other benchmark cities, the proposal continues. Thirteen percent of commuters bike or walk to work. There are 77 miles of bike lanes.
That scenario might not sound familiar to residents of Western Pennsylvania.
If you live in or near the city — complete with its quirky intersections, hundreds of bridges and a light-rail system that only connects to suburbs in the south — you know the average person doesn’t feel this way.
“If the Amazon people visited the city late afternoon, they would have got to the Greentree exit, turned around and went back to the airport,” Twitter user TC Clark wrote on Nov. 13.
Armed with a trove of data from ride-hailing companies such Uber while gathering insights gleaned from Pittsburgh startups working in the transportation space, researchers are collaborating with the city to come up with solutions that could sway even the infrastructure naysayers — and that’s happening regardless of Amazon.
The challenge? Integrating various datasets, said Alex Pazuchanics, assistant director of planning, policy and permitting at the city’s Department of Mobility and Infrastructure.
“Everyone’s data represents different levels of granularity or time increments or geographic coverage,” he wrote in an email.
In other words, some datasets have more detail than others or measure different variables or cover different communities.
“Still, all of the new sources are really valuable for rounding out the picture and helping planners make better recommendations,” Mr. Pazuchanics said.
Private industry, public insight
Startups that collect data about the region’s roadways should be especially helpful for the city’s infrastructure plans, according to Ryan Green, co-founder and CEO of Gridwise. The North Sidebased startup uses data to help ride-hailing drivers make better decisions that will ultimately help them earn more money.
“If we understand patterns across platforms, we can cause infrastructure change,” he said.
Gridwise is not working directly with the city, Mr. Green said, but it has engaged in exploratory conversations. He envisions the two working together to improve infrastructure and policies related to ride-hailing services.
For startups, a public-private partnership offers not only industry credibility — Mr. Green notes that once one or a few cities work with a given startup, others will follow — but also financial incentive.
Early on in a startup’s lifespan, there’s a great deal of market research to conduct, and the companies are not necessarily selling data yet. As Mr. Pazuchanics noted, “They may want to figure out what the data actually is that cities want.”
The city already pays for data services but is open to paying for other sources if it fits its needs, he added.
Plus, much of the data that these transportation startups are collecting is what Mr. Pazuchanics calls “data debris,” or information collected through the course of actions they’ve already taken but not central to the business. By selling that data to cities, it opens up another revenue stream.
A partnership with Gridwise could unlock new solutions, Mr. Green said, such as establishing optimal pickup and drop-off locations on side streets, creating a ride-hailing pickup zone that could improve the flow of traffic.
The company found that the average speed of ridehailing drivers is at its slowest on Fridays and Wednesdays between 4 and 7 p.m., meaning drivers are commonly stuck in traffic. That kind of data could empower the city to make decisions about bus routes during these periods.
Data also shows the Hill District is one of the least covered areas of trip activity within the city. That could inform policies in underserved communities or illustrate where there are gaps.
Other startup companies operating in Pittsburgh are also collecting data about the roads.
LaneSpotter, a free app that allows cyclists to crowdsource maps with the safest and most dangerous bike routes, is using the data it has collected to come up with ideas on where new bike lanes should be in a given city.
“Over the last few months, we’ve had a lot of input about the best corridors to put cycling infrastructure to improve safety or make a connection,” Mr. Pazuchanics said. “And I think LaneSpotter is another filter we could run the data through.”
RoadBotics, a CMU spinoff based in East Liberty, uses a cell phone mounted to the windshield of a car to analyze images of streets, looking for signs that a stretch of road will soon develop cracks or potholes. Municipalities then use the data to decide which roads will be resurfaced each year.
Even Uber has made anonymized data about driver trips available to the city through its Uber Movement tool. The data is meant to inform city planners, who historically have relied on sparse or outdated data.
Fusing the data
Combining this information about the region’s roadways has proven to be a challenge.
Usually such data is siloed and used for a specific purpose, such as when a city comes up with bus routes, explained Zhen Qian, director of CMU’s Mobility Data Analytics Center in Oakland.
And even though startups like Gridwise and LaneSpotter could enable better infrastructure decisions, the “oldfashioned way” is still popular, he said.
To predict traffic on the Parkway East, for example, city planners rely on inductive loop detectors, buried in the pavement, and radar detectors set up on the shoulder to count cars.
“It’s very useful, but it misses a lot things,” Mr. Qian explained. “It’s a fixed location number. You don’t know where people are going, where they’re coming from, what route they take, why they take Uber over the bus. This is what I’m missing.”
Not all of the city’s attempts to understand traffic have been old-school or analog.
The city and the Southwestern Pennsylvania Commission recently began using a new dataset called “StreetLight” that uses anonymized cell phone data to “understand how vehicles move through a designated area,” Mr. Pazuchanics said.
That could help the city understand how many left turns are made on one street compared to another — which could potentially lead to new turning lanes, eliminating some of those notorious Pittsburgh lefts.
Mr. Qian at CMU, meanwhile, is working to understand how data from public agencies and the private sector can be used in tandem to analyze travel behavior and improve infrastructure.
He and his colleagues have compiled data from ride-hailing services, bikers, public transit, roadway traffic, parking information and social media.
Lessons are even coming from last-mile delivery companies such as UPS and FedEx.
Truck flow has long caused congestion in cities, so companies like Amazon have put “micro depots,” like lockers, in strategic places to help deliveries to be streamlined and completed by a smaller fleet of vehicles in less time.
“We want to understand how people are moving in a network,” Mr. Qian said. “If we understand those flows, we can design both the infrastructure as well as how to manage the demand side.”