Univrses’ AI platform is using smartphones mounted in cars to ease congestion.
The world is currently experiencing a population boom of epic proportions, with cities bearing the brunt of overpopulation, and metropolitan areas undertaking the largest upsurge of urban growth ever experienced in human history.
Current statistics show that 50 per cent of the world’s population live in cities. The most conservative estimates expect that percentage to rise to by as much as 70 per cent by the year 2050. As overpopulation drives more people into urban areas, almost all of the difficulties currently facing cities will be intensified.
2050 may seem like a long way off but the effects of urban overpopulation are already making a considerable impact on daily life in many of the world’s cities. Stockholm is a prime example. While migration brings top talent to the Swedish capital, the population boom is also the primary cause of many of the city’s difficulties.
The current population of Stockholm is 967,000 inhabitants and 323,000 of those citizens moved into the city within the last 10 years. In real terms, that population increase is equivalent to the entire city of Malmö—Sweden’s third largest city.
As one of the fastest growing regions in Europe, the greater Stockholm metropolitan area has faced rapid urbanisation, and the rapid increase of the problems that go along with it. According to a report conducted by the KTH Royal Institute of Technology, published through the Stockholm Chamber of Commerce, some of the biggest challenges facing the city include housing shortages, a sharp rise in property prices, and overwhelmed public transport infrastructure. The report also highlighted the rise of workers commuting into the city to work from neighbouring municipalities. With unaffordable housing and with public transport overwhelmed, it’s the cities’ roads that are beginning to pay the price.
One problem that’s often overlooked when looking at population growth is the toll that is taken on urban roadways. Busier cities bring increased traffic, and increased traffic causes greater damage to a region’s road networks. Potholes develop faster, small cracks quickly evolve into larger ones, and every repair increases the cities congestion.
Though Stockholm is a relatively small capital city, it hosts more than 3,500 km of municipal roads. Estimates suggest the road network requires between 5,000 and 8,000 individual repairs per year. Currently, damaged sections of roads are reported by citizens via the city’s Tyck till app or by a hotline phone call. When damage is reported, each individual site requires a manual inspection. At present, Stockholm employs 12 dedicated road inspectors; a surprisingly small figure and clearly not enough to individually survey every single reported problem.
Population growth is taking a toll on urban roadways
While an individual pothole can be fixed in approximately 30 minutes, physically locating the problem can be time-consuming. And potholes are only a small part of the problem. Damaged, obscured, or outdated street signage is also a costly expense. Estimates suggest there are between three and five million traffic signs across the metropolitan area but a lack of official figures underlines the difficulty in maintaining the infrastructure.
Due to the limitations of manual inspection and the sheer volume of traffic signage, it can take up to 10 years before a damaged or unreadable sign is discovered. This includes reports made by citizens, but with a yearly average of 8,000 citizen reports concerning road signage, it’s not enough for the City of Stockholm to gather a clear picture of the state of its road signage infrastructure.
Physical descriptions can often be inaccurate and GPS tags can be incorrect. Even when a pothole or street sign is successfully located and repaired, it only presents a small snapshot of the damage to the city’s roads, as only a few concerned citizens use the app, and manually inspecting 3,500 km of road just isn’t practical. But one company has come up with a novel means of tracking the health of Stockholm’s streets.
Univrses is a Swedish 3D computing and machine learning company that has been working with the City of Stockholm to solve its road maintenance challenges. It was founded in 2015, originally creating headset tracking technology for AR and VR platforms. Now it is working closely with companies in autonomous driving, such as the Volvo-owned Zenseact and mobile robotics, such as ABB. Working with these industrial companies, Univrses has developed specialist perception technologies that allow vehicles to understand where they are in space, the 3D structure of the world around the vehicle, and even the exact nature of that 3D structure.
But the company quickly realised that its technology had wider use cases than solely within the autonomous vehicle sector. Jonathan Selbie, Univrses’ CEO, said its interest was piqued by a report from French multinational transport company Keolis, which found congestion and impassable roads were costing cities huge amounts of money. He says: “Working with autonomous driving already put us in the right domain to understand urban infrastructure and search for areas where the need for more data was greatest.”
Stockholm was one such city with its rapid urbanisation leaving gaping holes in its data sets. Geertje de Lange, a business analyst at Univrses, highlighted a prime example: “As cities are growing, there is a greater need for city planners to understand more about their assets. Most cities don’t know where their traffic signs or streetlights are.”
With most road inspections and inventory counts being performed manually, these time-consuming visual jobs can now be performed by smart cameras. Thanks to the advent of smart technology, the costs of such an operation, that relies on bespoke hardware, software, or labour-intensive operations, are no longer prohibitive.
As cities are growing, there is a greater need for city planners to understand more about their assets.
“Collecting data is not a new idea, but it’s how we collect the data that sets Univrses apart,” Selbie explains. “Most cities have data, but it might only be collected once a year. They can do this with scan cars [akin to Google’s Street View cars] or by aerial photography. They can collect this data but it’s expensive to get. It’s also a one-time effort, and it’s not up to date. There’s no way to spot trends. Our way doesn’t want to replace that, it wants to complement it by giving city’s access to real-time data.”
To do this, Univrses relies on smartphone technology. With a pre-installed app and a forward-facing mount, their 3DAI™ City program can scan and map the city streets from any car, transmitting and processing data in real-time. The app can identify anything that the human eye could see and analyse that information instantly.
Given that the City of Stockholm has its own fleet of approximately 700 vehicles, from garbage trucks to maintenance vehicles, it’s a cost-effective solution that’s easy to deploy and captures invaluable data.
“Parking occupancy is one metric that is better monitored by our system,” explains de Lange. “Parking is something that changes every hour, with significant differences between night and daytime occupancy.
Currently the city only measures parking occupancy twice per year using expensive consulting companies, and even then, due to limited resources, they can only cover 40 per cent of the inner-city streets, and only at one time per day. Even with a comprehensive study, it’s clear to see that the city can’t get accurate information about real occupancy.”
With Univrses’ 3DAI™ sensing capabilities, it’s possible for the city to use existing resources to help create a more detailed picture of the state of the city’s streets, measuring everything from road damage to congestion, the location of streetlights to the state of road signage.
Fleet vehicles are ideal for data harvesting and the benefits of decongested roads smooth surfaces extend beyond city hall. Taxi Stockholm is one company that has signed up to use Univrses’ technology to help streamline its operations, improve revenue, and reduce its carbon footprint. With 1,600 cars in operation around the city undertaking up to 22,000 journeys per day, Taxi Stockholm is an ideal data harvesting asset.
The 3DAI™ City platform can scan for roadworks and congestion to help taxi drivers deliver the fastest and most efficient service to their customers. The technology can also scan for spontaneous gatherings of people and direct empty taxis to the best locations to pick up fares. The result is a practical and efficient mobility service that keeps Stockholm on the move. Sustainability is also at the heart of Taxi Stockholm’s business model, with the firm striving to reduce emissions and decrease their use of fossil fuels. By using 3DAI™ City, journey times can be reduced, lowering emissions in the process.
“Taxi Stockholm wants to harness the power of advanced technology to deliver a premium service for our customers whilst responding to the needs of the city,” Taxi Stockholm CSO Andrej Smailagic explains. “3DAI™ City gives us the insights we need to do this. Based on early impressions, we believe that 3DAI™ City can help us become the most efficient taxi company in Europe and beyond.”
The use of 3DAI™ City has extended beyond Stockholm’s city limits too. Another early adopter of the new technology was Sweden’s transport association, known as Trafikverket. It is responsible for all transport infrastructure across the country, striving to operate and maintain Sweden’s roads in some of the most challenging conditions.
With a focus on sustainable infrastructure, Trafikverket partnered with Univrses to deploy 3DAI™ City technology with a view to reduce operating costs, prevent traffic disruption, and lower its overall environmental impact. The computer vision software captures data and processes it using sophisticated algorithms that can assess road quality or damage, highlight the location of temporary roadworks, or point out traffic signage that is no longer readable.
Meaningful data is extracted, processed and presented to the transport administration. With a clear insight into the state of the nation’s roads, it’s possible for decision-makers to quickly deploy resources and streamline their maintenance procedure to provide minimum disruption to road users. This is especially important in the far north of the country where roads serve as crucial artery routes, with few alternatives available if one closes due to road damage, extreme weather or other reasons.
In the future, Trafikverket is exploring how to expand the use of 3DAI™ City. This includes using the system’s traffic density and roadside pedestrian counting features to gain a deeper insight into both current and future problems that national road infrastructure may face.
Despite the positive adoption of the technology in Stockholm and around the country, any kind of mass-surveillance prompts privacy concerns. Naturally, all of Univrses’ technology is GDPR compliant, and while it may sound like the company is constantly filming the city streets, that’s not the case at all.
“We’re not filming the built environment but using our cameras to detect pre-determined objects. Our algorithms work on the smartphone “at the edge”, and are trained to look out for specific things, not everything,” de Lange explains. “As part of our service, we do capture images, but as soon as that data is captured and uploaded to the cloud, it must first go through an anonymisation pipeline that automatically blurs faces and license plates. No-one has access to that data—it’s anonymised before we even analyse it.”
Unlike static cameras that continuously monitor one spot, Univrses’ cameras are constantly mobile and can also be turned on and off by the operator at the touch of button. In that respect, and from a legal standpoint, 3DAI™ City is classed as a dashcam, unlike an intrusive static camera, and any privacy concerns have been explored through multiple privacy assessments.
No-one has access to that data—it’s anonymised before we even analyse it.
“3DAI™ City has been endorsed by several different organisations. These include Vinnova (a public entity also known as the Swedish Innovation Agency), the City of Stockholm, the Swedish Transport Administration and the City of Helsingborg. We are currently approaching the finish line of a full-scale Data Privacy Impact Assessment with the City of Stockholm, too,” Selbie added.
“Our 3DAI™ City service only collects data about specific objects, and not personal data. That’s why we’re keen to focus on working with cities and authorities, and we can align our goals with that of governments. We’re not watching what is going on, we’re sensing the city in a way that will reduce congestion, improve mobility, and save money, using computer vision, deep learning, and vehicles that are already out there.”
While road infrastructure might not be one of the sexiest aspects of smart city development, studies have shown that maintenance is an immensely expensive undertaking, and not just in Stockholm. For example, in 2016, the American Automobile Association said potholes cost US drivers $3 billion annually. In the Netherlands, roadworks caused more than two million hours of traffic delay per year, and every hour that a single vehicle is stuck in a traffic jam, it costs the government and taxpayer €10.58. It’s a similar story in London, which is one of the world’s most congested cities. Roadworks in the British capital causes congestion that costs the economy a staggering £4 billion a year.
Taking control of the city streets is a worthwhile endeavour, but the nature of the technology means that there are other practical applications to 3DAI™ City’s computer vision than counting potholes. In future, the company plans to integrate more features that can better serve city governments and citizens, such as providing detailed data about crowd volumes, parking occupancy, and even more unusual avenues, like counting windows for insurance or security purposes.
The ultimate goal of Univrses is to make things more efficient: to use resources more efficiently, to make better use of taxpayer’s money, and to make cities a more comfortable and convenient place to live. The challenges that Stockholm faces aren’t Stockholm-specific, but after using the Swedish capital as a test bed it won’t be long until Univrses expands and implements their smart solution in other congested cities around the world.
“We’re not a road monitoring company. We’re in the business of finding out where there is a clear need for data, and how we can produce that data in a cost-effective and scalable way,” says Selbie.