Why aren’t the potholes fixed yet?
It definitely caught the attention of the media when the ability to spot potholes more quickly was included as an example in the AI Opportunities Action Plan: government response.
Where I live, potholes are particularly bad at the moment. In recent years we’ve had more extreme weather, including flooding, that has caused damage to roads. As well as this climate impact, there are significant changes to road use, with heavier vehicles, including SUVs and electric cars, now more common. All of this means road surfaces are in worse repair and have more need of regular maintenance.
If you ask anyone local about potholes most people have stories. They’ll describe damaged cars or tyres, as well as offering opinions about how road surfaces aren’t being maintained, or whether temporary fixes are doing a good enough job or lasting for long enough.
This is also not a new issue. A quick search found this article in our local paper, with a headline from 2019 reporting that potholes in Cumbria had risen by 94 per cent.
Can AI solve the problem?
This post by James O’Malley is probably the best explainer for how AI can create efficiencies in road maintenance. James explains the use of relatively proven technologies which are becoming more powerful and increasingly accessible via different providers. It involves road maintenance vehicles being fitted with cameras designed for AI image processing, GPS receivers, and 4G internet connections:
“Together, this enables the road maintenance teams to automate the most annoying parts of their work. Instead of having to stop and get out of the car when they spot a problem, they can simply… continue driving.”
As James explains, some UK councils are already using this technology and the platforms being designed to work with this data. This includes how aspects of road maintenance management can then be managed effectively once defects or potholes are detected. It also opens up the possibility of imaging data supporting the work of other council teams, such as those needing to monitor fly-tipping in a local area.
We know that the technology described here will continue to improve. This article (Times paywall) talks about the latest developments in imaging AI, including the ability to identify road damage before it occurs. For example, distinguishing between potholes and the cracks from which they evolve so early maintenance can be prioritised. The team leading this work are currently building a ‘digital twin’ of the city of Cambridge where every road defect can be mapped. In the article they describe the following future vision based on the potential of harnessing more data from modern vehicles:
“Imagine a world where, rather than us inspecting the roads once a year or once every two years, as soon as three vehicles drive on top of a specific defect, it’s already detected and reported.”
It is a compelling picture, but one that is still detached from the realities and the real costs of road maintenance.
Understanding the cost of road maintenance
James Reeve shared a post on LinkedIn last month where he described how there’s no shortage of people submitting reports to councils through services like FixMyStreet. The RAC even have their own pothole index. James explained that:
“…while it only costs about 9p per report for councils like Surrey or Buckinghamshire that use FixMyStreet, the actual work to fix the roads is orders of magnitude more a cost of more than £50 per pothole (ref) – it would cost £14 billion just to fix the potholes that have already been identified […] AI won’t fix that. These are problems that will only solve with a drastic reimagining of how public services should be delivered.”
To be clear about the scale of the challenges here and how stretched local authorities really are, another Times article described the following as part of their visit to the UK’s pothole capital in North Wales (Times paywall):
“…councils are fixing a smaller and smaller proportion of the roads that need it. Figures from the Department for Transport show that, in the five years to 2023, maintenance rates dropped by 45 per cent. Separate DfT data showed that, of the roads assessed by councils as requiring repair, less than a fifth were mended last year.”
This article goes on to talk about how councils aren’t able to catch up, often having to resort to short term fixes or patches to roads which simple don’t last.
So, to go back to the point about how efficient it is to spot potholes. Yes, councils might be able to save some money with investment in AI. But perhaps most importantly, we also need to think about the cost and maintenance of the investment in new technologies here (e.g. imaging cameras, connectivity, etc). There’s the need for staff training and support, and the potential licensing costs of software and how this might be tied to future vendor lock-in or overheads due to a fixed number of suppliers owning solutions. This could all lead to increasing costs over time that could start to outweigh any efficiencies being created.
Solving the real problem of fixing the roads
With such a significant backlog, the cost and time required for maintenance is the key problem to solve, however there is already progress being made. This RAC article describes the recent launch of new JCB machines designed to fix potholes in minutes for the lower price of £30.
The RAC also reported on the world’s first road fixing robots. I would like to see how these cope on isolated country roads outside of urban or more controlled environments. This type of innovation also raises questions about how road maintenance is safely managed, such as how roads are shut and roadworks managed to minimise traffic disruption – something that still needs careful planning and human oversight.
But arguably the most important development in road maintenance is how companies are starting to innovate to improve the durability of road surfaces, including the science of developing self healing roads. This is the use of new materials that are less susceptible to cracking and pothole formation in the first place.
If we’re going to invest money anywhere, the science of how we adapt road surfaces so that they need less maintenance in the future feels like a strong bet, especially alongside any bets we make on imaging technology and automation to reduce overall costs. Otherwise, improving technology that detects road defects will simply continue to add to an already growing backlog of maintenance work needed.
Should AI be used at all here?
Finally, an important question with automating or removing any set of human interactions is: what might be lost as the result of this change? This is the need to think about the unintended consequences of making a change to how something like road maintenance is managed.
FixMyStreet was mentioned earlier. This was originally launched by MySociety in 2007 and has become an important service through how it engages local communities. It lets the public report potholes, but also many other things – examples given on their website site include: graffiti, fly tipping, broken paving slabs, and street lighting. It’s subsidiary company SocietyWorks works directly with local authorities and other public sector organisations to integrate digital solutions, including FixMyStreet Pro.
So what if the type of road maintenance efficiency described by the AI Opportunities Action Plan means that we lose something?
On the SocietyWorks blog in January, Angela Dixon, Managing Director, wrote the following in this post:
“There is plenty of research which explores citizen reporting and its impacts in relation to the fixing of local problems like potholes […] We have seen how much more likely people are to engage with councils if they can see that making the report makes a difference.
As the use of automation and AI models accelerates when it comes to the actual reporting of problems, it is so important that the public sector does not lose sight of the value in providing robust and community-centric reporting services, ensuring parity in the reporting process and enabling positive acts of citizenship to inspire further engagement.”
This emphasis on the importance of community reporting is also the ability for councils to work with communities in how they make decisions around the use of limited funds. It could be the chance to improve the overall infrastructure of a local area, also considering other factors like community health outcomes. As Angela goes on to explain:
“Councils are responsible for maintaining our roads, but by no means are they solely responsible for the potholes themselves.
As well as funding for fixing potholes, we need funding for improving public transport and cycling infrastructure, among other initiatives that would help reduce the pressure on our road networks.”
Although sometimes splitting local opinion, there are many other factors to what might reduce the need for road maintenance in the first place, including speed limits, creating more pedestrian friendly zones, and encouraging people to use their cars less with improved local transport investment or infrastructure for cycling.
But what could be more powerful here is the ability for future community-based services to be integrated with AI based platforms and new uses of imaging technology. This could mean a more open, transparent approach, where the public continue to interact directly with councils in a meaningful way, but while local authorities also realise the benefits of automation and find the most efficient ways of managing road maintenance.
In summary, there’s a lot to think about beyond the headlines of AI being used in how we tackle potholes. The reality is we’ll still be hearing the question “why aren’t the potholes fixed yet?” for the foreseeable future.
Related to this topic: Another good example of imaging technology is shared in this article published by Herald Wales. It describes how, by using advanced image capturing and AI data-processing technology, scientists have worked with environmental charity Hubbub to create a detailed map of the locations and types of litter dropped across the roads of Cardiff and South Wales.
This is my blog where I’ve been writing for 20 years. You can follow all of my posts by subscribing to this RSS feed. You can also find me on Bluesky and LinkedIn.