Robin Bennett runs a software company in Telford. He also drives around Shropshire, and he found the amount of rubbish dumped along the roadsides “really depressing”.
So he built a litter dashcam.
How it works
The Roadside Litter app runs on a phone mounted on the dashboard. “I’ve got a phone holder stuck to the dashboard, I’ve put my phone in securely,” Bennett told the BBC. “I start it and that’s it, you don’t have to touch the screen, and the screen will go black in a minute so it’s not distracting in any way.”
From there it is automatic. “Every 10 seconds it takes a photo, it looks at the litter it can see by the side of the road and it grades it according to a national standard.”
Three design decisions are doing the work there, and they are worth separating out.
The first is that it requires nothing of the driver after the trip starts — no tapping, no judgement call, no reporting form. The second is that the screen goes dark, which is what keeps it from being a distraction and, incidentally, what keeps it legal to use. The third is the one that matters most: it grades against an existing national standard rather than inventing its own scale.
The numbers it has produced
The app has logged more than 3,000 litter reports in Shropshire and more than 28,000 across the UK.
Bennett started it to answer a question he could not otherwise settle: whether the problem was particular to his county or part of something national. That is the honest version of a complaint — not “this is terrible” but “I do not know how bad this is compared with anywhere else, and I would like to”.
The data has already produced a finding. Litter was particularly common near convenience stores, fast-food outlets and motorway service stations. “People pick up a bag of fast food and just throw the bag out the window,” Bennett said.
That is not a surprising conclusion. It is, however, a located one — it names the kind of place rather than the kind of person, which is the difference between a grievance and something a council can act on.
What happens to it next
Bennett approached the campaign group Clean Up Britain to ask how he could help, and is now sharing the data with them in the hope it will encourage action.
Shropshire Council said it recognised the importance of maintaining a clean and attractive highway network, and was engaging with National Highways to explore opportunities to improve litter management on the A5 and A49.
Read that carefully: a council is discussing two specific roads. Whether the app caused that is not established, and the BBC does not claim it. But the mechanism on offer is clear enough — before the app, a complaint about roadside litter was an opinion about a feeling. After it, it is a count attached to a stretch of road.
Why this is a better AI story than most
There is no model here that decides anything consequential. It classifies rubbish by the roadside against a published scale, and if it occasionally grades a verge wrong, nobody is harmed.
That is precisely what makes it a good use. The AI is doing the one thing automation is unambiguously good at: performing a tedious, repetitive judgement at a volume no person would sustain. Nobody is going to photograph a verge every ten seconds for 28,000 observations. The machine will, and the output is a dataset rather than a decision.
The version of this that would work here
The pattern transfers to a great many Bangladeshi problems that are currently argued about rather than measured — illegal dumping, waterlogging on particular stretches after rain, broken footpaths, unlit roads, potholes on a specific route.
What makes Bennett’s version work is not the model. It is that he found a national grading standard to measure against, so his numbers mean the same thing as everyone else’s, and that he gave the data to an organisation whose job is to push on it. A dataset nobody has agreed how to read, handed to nobody in particular, changes nothing.
The technical part of this project is the easy part, and it was built by one annoyed person in his spare time. The hard parts were choosing a standard and finding someone to hand the results to.
