This interview was conducted by Freddie Price (Senior Partnerships Manager, Apolitical), as part of a series for Apolitical’s 50 States, 50 Breakthroughs, created in collaboration with the National Academy of Public Administration and Humans of Public Service.
The City of Boston’s Office of Emerging Technology is using modern data tools and AI to tackle a deceptively hard challenge: curb space is one of the city’s most contested public assets, but the rules that govern it are often complex, inconsistent, and hard to interpret, both for residents and the people managing them. Boston is building a digital model of its curb network using street-level imagery, LiDAR data and AI to translate physical layouts and parking regulations into accurate curb data. With this foundation, Boston’s Office of Emerging Technology, Parking Clerk and New Mobility teams will pilot dynamic curb policies that respond to real-time demand, improving safety, access and efficiency.
In conversation with Apolitical, the City of Boston's Office of Emerging Technology, including its Director Michael Lawrence Evans and Transportation Technology Strategist Sam Brenner, reflected on what drove this work, how they approached the underlying challenges, where they began, and what they've learned so far.
What problem were you trying to solve, and why did it matter?
Michael: The curb serves an enormous range of needs, from parking to deliveries to accessibility. And we were struggling with an extremely manual and inefficient process of tracking and managing it across the city. The information existed in fragmented ways. Sometimes it was just in people’s heads; other times it required our transportation planners to manually check Google Street View. When redesigning streets or allocating space for new uses, our staff had to walk streets inch by inch, documenting every sign to align the locations of that signage with engineering drawings.
This made it difficult to manage the finite resource of curb space and make informed planning decisions about everything we want to use the curb for: parking, delivery zones, outdoor dining, and other competing uses of this very valuable public asset.
Before this work, how did residents understand where they could park?
One of Boston's more notorious parking signs, in the North End.
Sam: There really wasn’t much information for residents beyond the signs themselves, and the signs can be incredibly complex.
What came up over and over again was both the complexity and the inconsistency of signs: inconsistency in how signs are put up on the street, and inconsistency in how they’re regulated. So much is neighborhood-specific, which becomes ‘tribal knowledge’ and creates a lot of friction and frustration for residents.
It's worth noting that when people have bad experiences at the curb, it really degrades trust in government. We've talked to a lot of residents who feel like the City is out to give them tickets or tow their cars; we think we can change that with public curb maps and better curb management. There's a huge opportunity here to do a better job making curbs easier to navigate and consequently make residents feel like the City is working for them and not against them.
When people have bad experiences at the curb, it really degrades trust in government... There's a huge opportunity here to do a better job making curbs easier to navigate and consequently make residents feel like the City is working for them and not against them.
Do you have an example of that friction?
Sam: We’re constantly conducting user interviews, and recently we were showing people a notoriously complicated sign in the North End. In back-to-back-to-back interviews, the immediate reaction was basically: “No, I’m not going to take the time to read this.”
Or people would spend a few minutes reading it and then say, “I’m not confident in my decision, so I’m going to move anyway.” Both outcomes are failures for the city. It creates circling, and it can hurt businesses. There are so many downstream consequences when people can’t make quick, confident decisions about where to leave their car.
Michael: When our regulations are that hard to interpret, it's a sign that the underlying system is broken. We have inconsistent data, conflicting rules that have accumulated over time, and no single source of truth. The sign is just where residents encounter that problem.
What did you build, and what are the main components?
Michael: We received a SMART grant from the U.S. Department of Transportation to digitize our curb regulations with machine learning, and we’re moving forward a portfolio of work under that.
At a high level, there are a few core pieces:
- Extracting rules from signs at scale
We’re building a tool that uses multimodal models to extract regulation information from images of street signs. We have tens of thousands of signs across the city, so automating that extraction makes a huge difference.
- Integrating with existing city systems
For example, we’re integrating with Cartegraph, our asset management system that stores information about installed signs. We’re also using Cyclomedia, which provides panoramic street imagery (similar to Google Street View) and a 3D representation of streets using LiDAR.
- Storing regulations in a standardized, machine-readable format
We’re using open standards (specifically the Curb Data Specification – CDS) so the data is stored in a way that’s immediately usable by other cities or companies, and easier to work with quickly.
- Building resident- and planner-facing interfaces
We’re creating an interactive map so residents and planners can understand parking availability and current curb use. Longer term, we want it to support scenario planning — for example, if you add a loading zone, or remove parking for a bike lane or bus lane, how does that affect congestion or emissions?
- Designing for re-use, not lock-in
A lot of cities buy off-the-shelf solutions that may be customized, but aren’t open source and can be subject to subscription fees. Our work is being built with vendors, but it’s still customized, and it’s intended to be open source. The hope is that other cities can follow along and reuse it, and that it becomes fairly plug-and-play because of the standard formats.
How did you design and validate your approach?
Sam: Curbs affect so many different types of users, and there are a huge number of city curb management processes, operational decisions, and policies. That put us in a position where we really had to prioritize which problems we wanted to solve first, and for whom.
The biggest early decision point was are we building for residents first, or are we building for internal City Hall users first?
We did dozens of interviews, we looked at 311 data, we did neighborhood walkthroughs with community leaders, went to coffee hours, and built an understanding of the players and the types of problems.
But the moment that clarified our approach was building a prototype map and putting it in front of people. Once we had that prototype, it became incredibly obvious that nobody was going to use anything we built unless the data underneath it was accurate, comprehensive, and reliable.
That answered a key design choice for us. We were deciding between focusing on interaction design first, or system design first. We started with system design, getting the underlying data right, and at the same time, that gave us runway to continue user research both inside City Hall and with residents.
Once we had that prototype, it became incredibly obvious that nobody was going to use anything we built unless the data underneath it was accurate, comprehensive, and reliable.
Michael: I would love to stress that our work to validate the approach cut across City Hall. In particular, the Office of the Parking Clerk, led by Mia Capone, is in the middle of a major digital transformation of Boston's parking infrastructure. Her team is modernizing payments, permitting, and ticket management, replacing a system that has been in place for at least 40 years. And transportation engineering, led by Amy Cording, is rolling out new asset management for Boston's 200,000 plus signs. They manage a high volume of curb changes and know where the system breaks down. We wouldn’t be able to build our tools without their expertise.
Why start in Chinatown, and what did you want to learn from the pilot area?
Boston's Chinatown faces many competing curb demands, which made it the natural starting point for the pilot
Sam: We chose Chinatown because we wanted the greatest impact possible as soon as possible. Chinatown has some of the most significant curb challenges anywhere in the city.
It has major congestion issues, a huge density of commercial deliveries because you have a lot of restaurants in one area, plus pickup and drop-off needs. And it’s gone through decades of redevelopment in a small area, which has left a patchwork of curb management approaches that don’t work well together.
There’s also a practical advantage that it’s condensed. One of the most significant parts of the challenge is getting accurate sign data. In Chinatown, you can cover a handful of blocks and get a huge range of cases to work with. It gives a lot of leverage for iterating and running experiments.
And for scaling beyond the pilot, there’s a lot of variability in Chinatown signage. If we can solve 80–90% of the edge cases there, we’ve probably solved 80–90% of the edge cases in the city.
How has collaborating with other cities in the SMART Curb Collaborative informed your approach, and what lessons from their experiences shaped your decision-making?
Michael: The SMART Curb Collaborative has been very important in shaping our approach. We met monthly with other cities across the country who received similar funding to improve curb use, including places like Portland, Minneapolis, and Los Angeles.
A big theme was procurement. Every project involves a lot of procurement, so we traded notes on challenges and on vendor experiences. Those cities had received their grants about a year before us, so we could learn from what they’d already run into, including how ready the technology was and how collaborative different vendors were.
The collaborative was organized by the Open Mobility Foundation, which oversees the development of both the Curb Data Specification and Mobility Data Specification. By the time we came into the collaborative, CDS was much more mature, and that made it easier for us to come in with an open-source, open-standards approach.
What’s something you wish you knew at the start of this project that you’d tell other cities trying to replicate this work?
Sam: First: success or failure comes down to the accuracy of your data. Solving curb management requires collaboration and sharing, and if you’re not collaborating and sharing accurate information, it falls apart.
Second: technology is exciting, but the thing that determines whether you can create and maintain accurate data isn’t the tech, it’s the people and processes. You need to rely on the people changing signs on the ground to update things accurately. You need teams that may not have worked together before to collaborate and share data in the same platform. None of it works without relationships and trust.
And one more: a lot of the technology work that moves the needle most on policy outcomes involves making connections across departments. Being positioned outside a single department can give you the ability to see gaps, connect dots, and find missed opportunities. That’s where you can get real ROI for taxpayers.
Technology is exciting, but the thing that determines whether you can create and maintain accurate data isn't the tech, it's the people and processes.
Michael: We initially thought we would hire specialized data science talent to do a lot of this work. But it's hard for city government, especially with grant funding and a limited timeline, to hire someone with that specialized skill set into roles that aren't clearly permanent. That pushed us toward procuring services and working with vendors, while making sure everything we built stayed open source and reusable by other cities. Unfortunately, procurement takes time and put pressure on our schedule.
Another challenge is remote collaboration. Working with vendors you’re not co-located with, across time zones, while they’re juggling other projects. If you have a team physically located together, the work can be much smoother.
Log in or sign up to continue the conversation