With driverless cars and electric hire bikes, our roads are going to look very different in the future. And if Richard Cartwright has his way, there will be far fewer traffic jams too.
The entrepreneur is working on a low-cost way of using data to free up roads. Every week, commuters waste valuable hours sitting in their cars, burning costly fuel and generating emissions and noise. It is estimated that drivers in London spend a dispiriting three days a year in rush-hour traffic, while commuters in Birmingham and Manchester lose a day. And it’s not just the inconvenience of it – the cost of congestion to drivers is estimated to be £30bn per year.
“We hear constantly from citizens and from traffic managers that congestion is awful and we want to do more about it. But cities don’t have the tools to do anything about it,” says Cartwright.
The entrepreneur left the University of Cambridge – a city with its own traffic woes – with a degree in economics and a desire to go it alone. After graduating in June 2017, he entered a business plan competition held by the Singapore Management University and won $2,500 for a proposal that advocated the idea of extracting data from CCTV cameras to manage traffic.
“It was a really cool opportunity to delve into the smart city concept,” says Cartwright. “We asked the fundamental question: ‘why aren’t cities doing more with their existing data?’”
Although it’s early days for FlowX, the startup he co-founded, it looks like he could be on to something. In October it was selected for the Geovation accelerator programme, run by Ordnance Survey (OS), that supports startups working with geolocation data. His business received a £10,000 grant from OS and office space in central London. He then set about visiting control centres across England, in London, Birmingham, Manchester, Leeds, Newcastle, Sheffield, Milton Keynes and Huddersfield.
Cartwright found that traffic data was surprisingly limited, rarely automated and not fit for the data age. He observed that a huge amount of traffic data is derived from magnetic induction loops under the road. “These are wires that note every time my car goes over them. I thought, ‘that doesn’t seem like the future.’”
He also found that spotting an accident can be difficult. Although there are cameras at most major intersections, there are so many screens to monitor that missing a collision is surprisingly easy.
“A major crash at an intersection creates major waves through the whole system. So [traffic controllers] really care about finding out about incidents as quickly as possible,” says Cartwright. “Traffic controllers don’t always see the accident, they just see the effects of the accident and, by then, it’s much more difficult to control. There is no computer analytics, no smart stuff, no computer vision on any of the CCTV in any of these cities. For me, that was pretty shocking.”
The answer, he believes, is the creation of a smart, analytical system to extract data from existing CCTV cameras. Computers could learn to identify and count vehicles on the road and understand what a normal flow of traffic looked like. Then, through machine learning, the system would learn how to spot an incident and to alert a traffic manager the moment it happened.
“Computer vision, combined with machine learning, has grown as a field considerably in the last 10 years, and it has made analytics on this stuff possible that wasn’t before.”
But Cartwright realised this was an undertaking he couldn’t carry out alone, and sought the help of London-based traffic software company Vivacity Labs, which already uses machine learning and AI to monitor and manage traffic levels. However it does this by installing its own sensors and cameras.
FlowX wants to combine Vivacity’s software with existing CCTV cameras to create a lower-cost solution. Cartwright will need to overcome the technical problem of integrating the systems. However it is the political problems, he says, that are a bigger concern.
While the system can work with anonymous data, privacy and personal data issues are a concern for the authorities, and the public.
“[CCTV cameras] were installed originally for traffic managers to look at. Now they would be used to anonymously count traffic information in the background. But as you can imagine city authorities are wary of that.” He is in talks with councils in Newcastle and Manchester and is bidding for a pilot project backed by Innovate UK in the former. He says councils are interested, but privacy and the use of CCTV is politically sensitive. “Manchester and Newcastle are initially open to this idea, but it’s quite daunting for them, I think,” says Cartwright.
If the project is able to take off, FlowX will be part of a new wave of traffic data companies providing analytics and tools to councils and traffic managers.
“My big vision is to enable any city, anywhere, to better use its existing information to manage that traffic and reduce congestion at a low cost. We want to enable cities to react faster to issues on the road. If you can do that, that should lead to real, tangible benefits to drivers, and everyone else.”