Car sensors fuelling a digital infrastructure economy

These days, modern cars are brimming with sensors making measurements about almost every aspect of the way the vehicle is performing, and - crucially - the condition of the road it’s driving over - how wet it is, how hot it is, even how bumpy it is, during every journey we make.

Now, at the moment, all this data resides with the car manufacturers. But researchers like Cambridge University’s Ioannis Brilakis are working on ways to marshall that data into forms that can be used to keep roads in better shape, going forwards.

This is the Digital Roads of the Future project, and the vision is to create a “digital twin” of the road network, updated constantly by data from drivers and used to spot the areas where potholes and other defects are beginning to develop and need attention before they become a major problem…

Ioannis - In the context of roads, we have a huge network out there of vehicles that have the ability now to collect data for us. We're getting to a point where the construction sector no longer needs to be involved in inspection. The car industry can do it for us quite effectively because they have the ability to capture video, accelerometer data, and several other modalities, including mini weather stations on the car. As a result of that, they can find the cracks, the potholes, the damaged lane markings, fallen street signs and other problems, and provide that information without us having to do a separate trip.

Chris - Who is doing this?

Ioannis - Mercedes is already doing this quite effectively. They have contracts in place with the Netherlands and Sweden and recently with TFL to provide such information, to the whole country in the case of Netherlands and Sweden. But the rest of the car sector is waking up to this and we can see Jaguar, Land Rover, Ford, and other businesses trying to get into that functionality.

Chris - Is this part of their business model then? Where previously their business was, they made a car and sold it to you and made money once and then made some money out of spare parts and a bit of servicing, now they've got this amazing sensor network on wheels. Is that where it's going?

Ioannis - Yes, I think Elon Musk showed the way when it comes to this by effectively positioning Tesla as a data company. All the other companies are trying to do the same because they understand there's a lot more value in using their own customers, the drivers, as data collectors rather than just simply selling the cars. This is what's happening. Mercedes is becoming a data company. They're now making a good part of their revenue through selling the data that comes through their customer vehicles, and this is only going to grow.

Chris - How much data are they collecting?

Ioannis - If I remember well, it depends on the car company, but from what I know from Mercedes, it's about 25 gigabytes per hour per vehicle. Of course, this is not all stored, only the useful extracted information is sent onto a Mercedes cloud and then Mercedes sections of that information into different categories and sells that as separate packages.

Chris - And is the world ready to receive this sort of information? Because that's one of the other projects that you've been working on here at the University of Cambridge, isn't it? That all this data can only be used if we've got a way of deploying it and interpreting it?

Ioannis - That's where we started with the Digital Roads programme. Going back to the example, we've used Mercedes, they are producing all this data, but our strategic road network with National Highways is simply not able to receive it. This is because of the growing gap between the technologies on cars in the car industry versus the technology on the infrastructure side. If we think about it, since the 1980s to today, cars have evolved from mechanical pieces to sensors on wheels, whereas infrastructure is still just as dumb as it used to be 40 years ago. Well, that doesn't fly. It doesn't work in the future. If we want to create modern infrastructure that's suitable for autonomous vehicles, we need to make the jump on the infrastructure side as well. Government can help by creating the kind of data set that can help all the car companies at the same time: digital twins. The biggest value of digital twins is forecasting and forensics. Looking back in time and looking forward in time. As I'm driving my car, I want to anticipate, predict, understand what might happen in the next five seconds, ten minutes as I drive ahead. If the car is reporting to the cloud that the road is slippery and it is wet, then the cloud can look at that information, run a simulation, and say that, given this information, you need to slow down, otherwise you have a very high chance of an accident.

Chris - So will your ultimate product be something that can marshal all of this data, put it into a format that then anyone can use in the way you've been describing? You are the sort of middleman here?

Ioannis - I could be wrong, but if I recall correctly the dataset that National Highways currently has has a value of well over 40 or 60 billion pounds. We need to unleash the power of that data set. We need to get to a point where this is not all hidden for the use of very few people, but it's actually public. Like the infrastructure, the physical infrastructure is public for everyone to benefit. That's what we're trying to enable in our use case following the maintenance path. We're trying to leverage the value from the inspection, from the cars, into the digital twins, and make predictions based on the information we receive. For example, if we have a really large pothole that's a safety risk, we need to fix it right away. Whereas if it's a little crack, that's something we can then put in next year's maintenance schedule.