For the last few years, Atlanta’s city government has been on a quest to utilize the latest technologies across IoT, big data analytics, machine learning, and artificial intelligence to tackle the city’s biggest challenges — everything from traffic to water to public safety and crime. This vision is called the Smart City, and it’s being pursued by metropolitan areas around the world to streamline efficiencies in the often-snarly business of running a large city.
However, Atlanta is taking it a few steps beyond even that. They’ve recently opened a Smart Corridor along North Avenue, a major in-town road, which will see over 100 IoT-connected sensors, research and data-collection to inform things like a future of autonomous and connected vehicles, and a partnership with Georgia Tech to analyze all of that data. And Atlanta’s Chief Information Officer Samir Saini says the multiple SmartATL projects, while right now operating on a limited pilot program-basis, are being actively studied to determine which will be deployed at scale across the metro area.
The city will officially kick off the Smart Corridor project with an open house event, Experience SmartATL, this week. But Saini says the city is trying to be even more forward thinking with their intentions for SmartATL; he says the data-driven decision-making across the city will eventually go much farther than municipal government. He intends for the data to live — and be used by private business, startups, and the public — on an open data platform accessible to all.
What have been some of the big Smart City success stories that have been deployed thus far?
Now keep in mind, across the board none of the solutions we’ve deployed thus far none are city-wide. Much of these are still demonstrations or small pilots at different stages of maturity. But even from the pilot ones, I can share some where we’re already realizing the benefits. So let’s start with water, smart water.
The problem that we’re tackling in water utility is wastewater overflow. We have a combined sewer system and overflow events are obviously pretty horrible. The pilot project was to deploy sensors in the manholes so that we could monitor flow levels, track that flow data into an analytics platform, and see a visualization that our watershed operations team can monitor and determine through the dashboard where we see a potential spill. This is already deployed across a few hundred manholes in areas we see the most overflows, and we’re already seeing the benefits — a reduction in the number of wastewater overflow events because the watershed team has this real-time or near real-time data.
[Smart City water project led by Commissioner Kishia L. Powell and the Department of Watershed Management]
Describe another pilot.
Another one that I can share is for transportation. We have deployed on North Avenue an adaptive traffic signal control system that combines AI with traffic queries. It’s an AI-driven traffic control system, which is pretty rare in any city. It uses a number of different technologies — for example, a thermal imaging camera at each intersection, so based on heat it detects it can determine how many pedestrians, versus bicyclists and vehicles. Then based on that data, it determines what to adjust the signal timing to and then relays that information to adjacent traffic lights. And if that proves to make a significant impact in improving travel time delay and flow, we’re going to go all-in and deploy it in a number of other areas of the city; particularly event spaces where traffic is a nightmare. So that’s going to be a big win.
That’s a company called Surtrac, a startup, based in Pittsburgh. Started by a couple of Ph.D. grads from Carnegie Mellon, with this technology that we hadn’t seen this really anywhere.
The other thing is it’s not just about the adaptive signal control system; it also can enable traffic signal preemption, so we can then deploy a sensor on our emergency vehicles and the light would automatically change when there’s an emergency vehicle coming. The technology also creates the basis for what would be needed for autonomous vehicles. So that’s also pretty exciting because we’re laying down the foundation for what we see as the future of mobility once we have more connected cars, autonomous cars, in our city.
[Surtrac traffic signal project led by General Manager Faye Dimassimo and Renew Atlanta]
What would it take to deploy these technologies from a pilot to neighborhoods across the city — to scale them?
We have multiple demonstrations happening across Atlanta — not all of these are just on North Avenue, that just has the highest concentration, but we have other demonstration areas. For example, our partnership with Georgia Power and GE and AT&T where we have installed GE’s smart streetlights. We’ve deployed several hundred across the city, with a high concentration in high-crime areas. The streetlight has a sensor on top of it that has a camera for video, an audio sensor, and an environmental sensor that can effectively smell what’s going on, to detect things like carbon monoxide levels.
[Smart City project led by Commissioner William M. Johnson and Department of Public Works]
We’re going to be testing use cases — one of the immediate cases is gunshot detection. If a gunshot occurs in a high-crime zone, the audio sensor can triangulate the precise location of the shots, even determine the type of gun that was fired and the caliber bullet. That information in real-time could be sent to the 911 center to immediately dispatch an officer to the location. Today when a gunshot occurs, we wait for 911 — and often nobody even calls so we don’t even know how many shots are being fired across the city.
We’re really excited about public safety — we also recently formalized a partnership with Georgia Tech. Over the past several months we have been developing a machine learning algorithm that can read and scan through police case reports, thousands and thousands of historical and active reports, and determine the probability of correlation across them. This is so challenging today because, first off there’s too many reports for any human to manually correlate. This machine learning algorithm uses natural language processing (NLP) to scan through, extract words from the pretext, understand the context in which those words are being used, across thousands of reports to calculate what the probability of correlation is between any of them.
What we’re finding while training the algorithm is that it’s already correlated crimes that we had to correlate manually from years ago. And those took forever for us to correlate, but the machine learning algorithm did it in a second. This is a one-of-a-kind thing that we believe no other police department in the country is utilizing.
Many of these technologies are created by new companies — so the City is becoming a client for startups. Do you think that over time this will help bolster startup activity?
Yes, absolutely. We partnered with a bunch of companies — some are big, some are small. We’re deliberately trying to find local companies because we get the double benefit of improving metrics along with promoting economic growth. So we’re excited about that.
Are there any particular cities you’ve looked at as a model for the whole Smart City program in Atlanta? Have you taken any lessons learned from anywhere?
I don’t think any one city has it all figured out. There are plenty of cities that have deployed smart technology — you’d be hard pressed to find a city that hasn’t done something, at least, in one or several areas. That’s actually easy now. But the smart city programs have gone on long enough where now, the bar is raised.
You’re not a smart city if you deploy siloed IoT or big data technologies. You are if you’ve made progress on building what we’re doing — a City-managed enterprise data platform where we pull the data from all those smart solutions and utilize that platform to to go to the next level, and more importantly take steps to democratize that data. That is the new bar for what it means to be a smart city. This is the evolution right now.
What will that city data platform look like?
We’re building the data lake and beginning to inject all the data from all these IoT sensors and devices into the cloud. We recently set this up, so we’re building very quickly every day, every week that goes by it gets bigger. We’ve already made a whole lot of progress.
Today cities deploy platforms and open datasets in different forms, really from the perspective of driving transparency of government. That’s great, but this is well beyond that. This is transparent data, but throw on top of that a city SDK and APIs that can be called by the business community and the civic tech community to build apps and solutions.
It’s not enough for us to just to employ smart city solutions and then just walk away — that’s not going to cut it. It’s our responsibility, if we’re taking seriously our mission, to take it a step further to collect the data from all the files and put them to use for different groups in different ways.
How can the platform, and democratized data, be used by private companies?
A good example is a project which is intended to reduce accidents within Atlanta. Atlanta was selected amongst two other cities by a consortium of companies called Together For Safer Roads, including IBM, AT&T, UPS, and others. A team from IBM has been working with us to build a machine learning algorithm that can determine the predicted collision risk on every segment of every street. They’ve built a dashboard for us to access so we can see what the collision risk looks like on our roads every day. Sounds great, right? And it is — we now have the ability to be more intelligent about how we adjust traffic signals, for example, or we can give it to the police department so they can intelligently dispatch officers to areas where accidents have a high likelihood of occurring. That’s all great.
The problem is, none of that actually reduces accidents. It improves our ability to respond to them, but how do we actually reduce accidents? The only way to do that is to take the data from this solution and make it available as a democratized tool, as an API, for the business community. If we did that, then Google Waze, Apple Maps, all the navigation apps could connect to that API, so a driver using these navigation apps or in a connected car can be notified by their smartphone or their car of the probability of a collision in real-time. Then we may be able to actually avoid an accident.
It’s not enough to just build a solution for the city to use. Democratization of the data across all these solutions is so critical in ways that we understand, and honestly in ways we don’t. We don’t yet know the value, the potential, of mixing up all the data from all these sensors in waste, water, transportation, safety, and in analyzing it holistically. There could be things that we can’t even imagine we could do with that data — and that’s the really exciting part about smart cities. It’s utilizing it to connect dots we can’t even conceive of connecting today, that could have a profound impact on improving quality of life.