Thanks to the relative ease with which local governments can now gather real time data, combined with the capabilities of artificial Intelligence, cities are realizing interesting new ways to run more efficiently and effectively.
Improving cities is a pressing global need as the world’s population grows and our species becomes rapidly more urbanized. In 1900 just 14 percent of people on earth lived in cities but by 2008 half the world’s population lived in urban areas, and the rate continues to grow. There were just 83 cities on earth with more than one million residents in 1950, while as of last year there were 512 such cities. In the United States, 3.5 percent of the land now holds 62.7 percent of Americans.
This article will look at how governments and companies are using AI right now in cities. It will mainly focus on three major categories of applications:
- Helping officials learn more about how people use cities
- Improving infrastructure and optimizing the use of these resources
- Improving public safety in cities
We’ll conclude with some of the future implications of these (above) smart city technologies and trends.
The first step in a city becoming a “smart city” is collecting more and better data. There is an old saying in the world of economic modeling: If you put garbage in, you get garbage out. If an organization doesn’t start with good data, trying to make predictions about how new government policies will work can end up deeply flawed or even counterproductive. Helping cities gather and process data is one place AI is currently being put to use.
AI Learns How People Use Cities
Cities have wealth of possible data sources, such as ticket sales on mass transit, local tax information, police reports, sensors on roads and local weather stations. One huge source of raw data that AI pattern recognition technology is making significantly more manageable is video and photos. NVIDIA predicts that by 2020 there will be 1 billion cameras deployed on government property, infrastructure, and on commercial buildings.
That is far more raw data than could ever be viewed, processed, or analyzed by humans. This is why only a small fraction of cameras are ever actively monitored by people. This is where deep learning comes in. It can count vehicles and pedestrians. It can read license plates and recognize faces. It can track the speed and movements of millions of vehicles to establish patterns. It can process the huge volume of satellite data to count cars in a parking lot or track road use.
To help cities handle this torrent of video NVIDIA launched Metropolis, their intelligent video analytics platform. NVIDIA has over 50 AI city partner companies providing products and applications that use deep learning on GPUs.
Similarly, AT&T launched their Smart Cities framework in 2015 and formed alliances with Cisco, Deloitte, Ericsson, GE, IBM, Intel, and Qualcomm. Earlier this year AT&T announced a deal to be the exclusive reseller of GE Current’s intelligent sensor nodes for connecting cities — a significant deal since GE just announced it will provide San Diego with largest smart city Internet of Things (IoT) sensor platform. There are a huge volume of applications of AI with IoT (some of which we’ve covered here on TechEmergence), but few opportunities match the massive scale of smart cities.
GE is going to start by installing 3,200 CityIQ sensor nodes throughout the city and may expand that to over 6,000. The data can be used to identify parking spots for drivers, help first responders, and identify dangerous intersections.
A straightforward application of machine processing video for cities is license plate recognition (LPR), which is used in numerous ways. For example, since 2014 the City of Galveston has been using the company gtechna’s Pay by Plate number parking system. Instead of traditional analog meters, people can pay by phone and LPR technology verifies who is parked legally.
(Readers with an interest in urban machine vision technologies should probably read our full article on AI applications for satellite imagery, which features a wealth of representative use-cases.)
The system did not require installing any new infrastructure at the city’s historic Seawall, and it reduced deployment costs while preserving the area’s aesthetic. The company claims in the first year the system produced $700,000 in new revenue and the city saw an 80% increase in collection rates.
Similar parking systems are in use in multiple cities and by large private institutions, which are almost mini cities unto themselves. Stanford University uses VIMOC Technologies’ LPR system in its parking lots to quickly check if vehicles parked there have the right permits.
AI Optimizing Infrastructure for Cities
A large amount of existing public infrastructure is underutilized, overused, or used inefficiently due to a lack of real time information among individuals, companies, and government agencies. Drivers don’t know where parking is available, according to analysis by INRIX New York City drivers spend an average of 107 hours per year looking for parking. Passengers not knowing how long a bus will take have lower ridership than buses they can track. Cities for the most party don’t know what is the right length for a stop light at every given minute. These are issues companies and government are addressing.
Anyone who has spent 10 minutes driving in circles in a city trying to find a parking space should be aware of this problem. It is a waste of your time, and circling around increases downtown traffic, wasting everyone else’s time. It may seem like a minor inconvenience, but multiply that by millions of people each day in hundreds of cities, and it adds up to a significant waste of net resources.
Well known companies are trying to address this issue from the individual’s perspective. For example, Waze, acquired by Google back in 2013, uses its network of drivers to provide real time data about traffic and accidents to help individuals optimize their routes. Cities are also trying to improve the situation from their end (we covered the specific AI use-case of Waze in our popular “Everyday Examples of AI” article earlier this year).
Smart Parking Garages
VIMOC is using AI to make parking easier in Redwood City. The company installed vehicle detection and reporting in two of the city’s large parking garages. The amount of available parking is displayed outside the garages on large LED signs and shared with an open platform for use by app developers. It provides the immediate benefit letting individuals know where parking is available, and in the long term, the wealth of data collected will allow the city to make planning and pricing decisions.
Adaptive Signal Control Technologies
Adaptive Signal Control Technology allows traffic lights to change their timing based on real time data. According to the Department of Transportation (DOT), “On average [Adaptive Signal Control Technology] improves travel time by more than 10 percent. In areas with particularly outdated signal timing, improvements can be 50 percent or more.” Given that traffic congestion costs the country $87.2 billion in wasted fuel and productivity (according the the DOT), there is a strong reason why numerous cities and companies are deploying this technology around the country.
The benefits of this technology where it has already been deployed have been promising. For example, San Diego installed 12 Adaptive Traffic Systems along one of its busiest corridors last fall and found they “reduced travel time by as much as 25 percent and decreased the number of vehicle stops by up to 53 percent during rush hour periods.”
Similarly, in 2012 the company Surtrac first deployed intelligent traffic signals at nine intersections in downtown Pittsburgh. It reduced travel times by more than 25% on average and wait times by 40% on average.
As a result Surtrac has been expanded to 50 intersections in the city with more planned. Of course these results come from the technology first being deployed on the busiest roads where it was assumed they would have the biggest impact.
Connected Public Transit Technology
This technology allows buses and trains to communicate with each other and the general public. Letting individuals know when buses or trains are coming and if they are going to run late makes them more useful to individuals.
The Massachusetts Bay Transportation Authority was the first agency to make bus locations and arrival-time predictions available, allowing developers to create tracking apps. Research on the impact of real time information on bus ridership in New York City found that it increased weekday route-level ridership by 1.7%.
AI Improving Public Safety
Smart cities aren’t just about reducing commute times and saving on fuel. The same networks of sensors and cameras are being used to save lives and fight crime.
The same LPR technology used to track parking is used by law enforcement to find stolen cars and track criminals. By 2014 LPR was already being used by an overwhelming majority of local law enforcement.
The same intelligent traffic lights normally used to improve traffic flow are utilized by ambulances and fire trucks to get to the scene of an emergency quicker and more safely.
Shotspotter, a company that automatically locates gunfire based on a sensor network, has its technology embedded in GE intelligent street lights. Last year Shotspotter alerted law enforcement to 74,916 gunfire incidents. In its recent IPO it received aggregate proceeds of approximately $35.4 million.
The growing body of data gathered by cities is being mined to find out which intersections experience accidents, and importantly exactly how these accidents take place, to prevent them in the future. As part of the Vision Zero effort to eliminate all traffic fatalities, cities are turning to big data to plan and prioritize infrastructure projects. For example, Microsoft and DataKind recently partnered with New York, Seattle and New Orleans to use data science to improve safety.
Highly connected cities also provide the ability to give individual hyper localized warnings about possible natural disaster. Do to its unique geography rainfall can vary significantly throughout the city of Seattle. To address this they created RainWatch which combines radar data with a network of rainfall gauges to monitor rainfall with a high degree of resolution. It allows city maintenance workers to more quickly respond to possible problems and provide more accurate flood warning to residents.
Concluding Thoughts on AI for Smart Cities
The grand long term vision of smart cities is full interconnectivity: Self driving cars, trucks, and buses all talking with each other as well as with smart highways, traffic lights, and parking garages. The whole system will working together to move people around with an incredible degree of efficiency and safety. A highly connected system that will save lives, save time, and save fuel. A reality that will be made more possible as the federal government moves towards requiring vehicle-to-vehicle communication build into new vehicles in the coming years.
It is also to provide engineers and city planners with an incredible wealth of data that can be used to promote safety, health, and economic growth. Right now researchers often rely on rough estimates of how people are using most roads and bike paths, but in the future they could have access to a minute by minute breakdown of every block.
The important thing is that we don’t need to see any new technology developed to see massive gains from cities becoming smarter. We have existing technology proven to be capable of improving parking utilization, safety, and significantly improving traffic; it just hasn’t been widely deployed yet.
As a result, the potential for growth for these systems and the companies that make them is significant. It is not surprising, then, that numerous large companies (Siemens, Microsoft, Hitachi, others) have put an increased focus on smart city technology.
Beyond the industries directly providing these services to local governments, the spending on smart cities could impact a range of businesses. Reduced traffic would mean cheaper shipping and technicians being able spend more time at job sites and less time moving between them. Fewer accidents could result in lower insurance costs for everyone.
Image credit: Phys.org