OPINION: Why accurate smart city data is key for the future of our roads


Mark Nicholson, CEO and co-founder of Vivacity Labs – a firm that uses AI and computer vision approaches to gather better data for real-time optimisation of the road networks – offers his view on why the right data is always the right answer.

There is an assumption that all data is created equal. This is simply not true. History is littered with technologies which provide “data”, but if you cannot trust the insights it is worthless. When it comes to city infrastructure, authorities who are making decisions on the future of our roads and cities are rightfully sceptical of accuracy and reliability claims that are provided without evidence, including how data is verified.

With humans producing 2.5 quintillion digital data bytes daily, it has never been more important to consider the accuracy, breadth and robustness of data in informing decisions that have an impact on the way we travel. Computer vision technology is providing insights that contribute to the evolution of smart cities, helping to promote clean air and active travel initiatives and make cities better connected. But if the data isn’t reliable, these projects will undoubtedly fail.

Quality = accuracy + breadth + robustness 

Cities need to harness technology that provides detailed, high quality insights, especially when it comes to our roads. With authorities facing tough decisions on the future of high profile schemes and their success, data allows them to make informed choices about what is working, or not. By adopting new techniques such as computer vision, authorities can derive a vast range of detailed insights, all from a single device. For example, when it comes to traffic sensors, anonymous data is collected on a range of classifications of road users, from joggers and e-scooters to cars and buses, as well as the behavioural analysis of each. Authorities can then analyse which schemes are working and where improvements need to be made to promote sustainable travel and clear air in our cities. This is only possible if the data they are working with is accurate and as high quality as possible.

However, as authorities migrate towards AI & computer vision, there is also a hidden pitfall to consider. Just like the magnetic loops, pressure sensors and radar systems of the past, accuracy and reliability are fundamentally important. Authorities must be able to differentiate between different providers and know that not all computer vision is the same.

Ensuring data is accurate

It is important that the data used to make informed decisions is as accurate and reliable as possible. Feature-level validation (where providers can test technology) and site-led validation (which ensures that the individual installation is high quality) both help to ensure that data is accurate. Equally, the data that authorities are presented with needs to be gathered from widespread conditions and over extensive time periods to give authorities the genuine insights they need. Once that data is collected, ongoing system checking should be carried out to prove its accuracy, and it should also be independently verified by an impartial organisation.

Once trusted, accurate insights derived from data can bring about dynamic and positive adaptations in the live environment. For example, variable message signs can provide live updates to road users in cities. Street lighting can be changed based on the needs of road users. By identifying near misses between different classification types and controlling traffic junctions, AI can deliver the best possible rate of traffic flow and have a hugely positive impact on road users.

Measuring what matters

As well as collecting data at a range of times and in different conditions, it is important to measure what matters. Counting emerging transport types – as well as gaining deep insights on traits such as speed, occupancy, congestion, turning counts and journey times – all help to create the most informed view possible. Good smart city data isn’t centred on one of these points, but must factor in all of them.

Equally, the future of the smart city has to be citizen-centric – it’s important to keep data safe and anonymous. For any potentially invasive technology, no personal data should be recorded. In the case of computer vision, the sensors, using AI, can detect what type of road user is passing in real time and feed this back as anonymous numerical data. For example, five counts of cyclists, two pedestrians and one bus. It should be a data feed – not a video feed. And if it isn’t, it jeopardises the trust of the citizens at the heart of these schemes.

Modern technology such as computer vision has the power to make our cities genuinely smart and can deliver the sustainability, connectivity and safety benefits that have been spoken about for years. In taking steps to go digital, authorities will need to be able to differentiate between the huge range of technology on offer and to prize accuracy and reliability above all else. Once the right technology is in place, the possibilities are endless.



Share this story:

About Author