Pipes or pennies

Bergen buried steel pipes under its historic centre so rubbish never touches a truck. Seoul just put a scale under the bin and a chip on your key ring. Both count as "smart waste management." One cost well over a billion Norwegian kroner and still can't say what it saved; the other cost the price of a kitchen scale and cut food waste by a quarter in its first trial. This article checks what genuinely works in municipal waste, at what price, using field measurements wherever they exist, and saying so plainly where they don't.

Half-empty trucks, fixed schedules

Most cities still collect rubbish the way they did in 1960: a truck drives a fixed round on a fixed day, emptying every bin whether it holds two kilos or two hundred. The waste is invisible until the truck is already there, so the only variable a planner controls is how often to send it, and erring safe means erring often. Multiply a slightly-too-frequent round by a whole city and the waste isn't just in the bins.

The obvious fix is to make the bin report its own state: a sensor that says "I'm full" or "I'm not," so the truck goes where it's needed instead of where the calendar says. It's the kind of idea that sounds too simple to need proving. Which is exactly the problem this article ran into.

Prague put sensors in 7,000 bins — and nobody published what it saved

The clearest run at this in the atlas is Prague's Golemio waste sensor mesh. A 2019 pilot fitted 464 underground containers with Sensoneo ultrasonic sensors reporting fill levels six times a day; by 2020 it was routine operation, and the network has since grown past 7,000 containers citywide, one of the largest fill-sensor deployments anywhere. It was also cheap to start: the 16-month pilot rented its sensor hardware for roughly 2.6 million Czech koruna across the 464 units, about €220 a bin.

Rome made a smaller, later bet on the same idea: from March 2024 its municipal utility AMA began installing solar-powered auto-compacting bins whose sensors alert crews through an app once a container hits 80% capacity — 68 of them across Testaccio, Trastevere, San Giovanni and other central districts. Both cities describe the same mechanism in the same confident language. Neither has published a number.

Not a rough number, not a range: nothing. Prague's own materials describe the mesh "cutting empty or overflowing pickups and the truck trips, emissions and noise they cause" without ever quantifying the cut. Rome's launch announcement counts the hardware — how many bins, in which districts, at what fill threshold they call for a truck — and stops there; what changed in trips, fuel or cost afterwards is simply not part of the story anyone tells. Two real, growing, multi-year deployments, and the industry's own operators haven't audited what either one is worth.

It gets more interesting once you go looking for who has measured it. A 2023 field trial by researchers at Lisbon's Instituto Superior Técnico and the University of Coimbra, run with the real Portuguese waste operator ERSUC on an actual Coimbra street, fitted four recyclable bins with sensors and compared sensor-triggered collection against the existing fixed schedule for seven months. The result: 32–47% fewer collections, depending on the fill threshold used, with overflow incidents down as much as 79–83%. It's real, it's peer-reviewed, and the authors call it themselves "a first exploratory test" on four bins on one street, explicitly warning against generalising it. A comparably rigorous, comparably real field study in Valencia found the opposite: bins never crossed 70% capacity under the city's already-frequent schedule, so sensors weren't economically justified there at all.

Meanwhile the numbers that do circulate widely turn out to be thinner than they look. Sensoneo, the vendor behind Prague's own sensors, reported "60% CO2 reduction" and "20% efficiency improvement" for its EU-funded SENS project to the European Commission, with no disclosed methodology, an official-sounding channel dressing up what is still a vendor's own claim. And a widely cited Lahore, Pakistan study with genuinely enormous data volumes — 180 million-plus sensor readings, 28 million GPS points — turns out, on close reading, to report the output of a routing algorithm run afterward on the collected data, not an actual measured comparison of trucks driving a real baseline versus a real optimised route. That distinction hasn't stopped the same handful of percentages recirculating, uncredited, across dozens of vendor blogs as if each repetition were independent proof.

The one place a number survives contact with reality

Rodgau, a commuter town of 47,000 outside Frankfurt, has the rare real figure, and it comes from a slightly different device: solar-powered "Econ Top" bins that don't just sense fill level, they compact waste to about a seventh of its volume before reporting in. Installed at busy spots like Puiseauxplatz from 2017, the bins let the town's utility go from emptying two or three times a week to roughly once every six weeks at the same locations, a reduction on the order of 90% in truck visits at those specific spots, self-reported by the operator, for one kind of location, never independently audited. It's the honest exception that proves how rare honesty is in this category: a small town with a handful of bins published more than global sensor deployments serving thousands of them.

The cheaper wins nobody's arguing about

While sensor-and-truck logistics struggles to prove itself, the plainest lever in this whole survey, price, has some of the best evidence anywhere in the atlas. Germany's bottle and can deposit scheme returns an estimated 96% of containers, a figure from the federal environment agency, independent of Deutsche Pfandsystem GmbH, the industry body that runs the scheme day to day (which claims a similar 97% itself). Norway's Infinitum system reported 92.3% returned across 1.54 billion containers in 2023, official but operator-sourced, and I couldn't independently corroborate it within this piece's research budget. And Seoul's weight-based food waste fee, RFID kiosks that weigh what a household throws out and bill per kilogram, measured a 25% average cut in food waste across eight pilot sites before the mechanism went nationwide in 2013.

The extravagant end: pipes under the old town

Bergen's Bossnett is the system that started this article. Approved in 2008 and switched on in October 2015, it sucks rubbish through buried steel pipes at up to 80 km/h from street inlets across the historic centre to two collection terminals, built by the Swedish firm Envac for BIR, the municipal utility owned by Bergen and its neighbouring municipalities. By 2018–19 BIR said more than 10,000 properties had access, with close to 100% system availability; the network keeps expanding toward a stated (not yet reached) target above 30,000. It is a genuinely bold piece of infrastructure, built specifically to pull collection trucks and fire risk out of a UNESCO-listed wooden wharf district that has already burned down twice.

What it isn't, after a decade of operation, is measured. No official or independent source publishes an emissions figure, a truck-trip reduction, or a clean total cost for Bergen specifically. Even the capital cost is disputed: an original 2008 estimate around NOK 1.05 billion grew to roughly NOK 1.5 billion by 2013, against an older, unreconciled reference to NOK 660 million for full build-out. At a 2018 public debate, BIR said the underground pipes would last "nearly 100 years"; a named local critic in the same debate put it at ten. Neither claim is resolved in the public record.

BIR says the pipes will last a century. A local critic, in the same public meeting, says ten years.

— Bergen's Bossnett, a decade in, still arguing about its own arithmetic

The broader technology has better evidence elsewhere, just not attached to the flagship case. A 2023 peer-reviewed review of pneumatic systems worldwide puts average investment at roughly €2,400 per dwelling, and cites a New York City feasibility study finding that pneumatic collection costs 3.3 to 6.6 times more in capital than truck collection for the same coverage, while running about 30% cheaper day to day. That's the honest trade at the heart of every pipe network: an enormous upfront bill against a genuinely lower operating cost, which is exactly the kind of arithmetic that makes "when does it pay back" a live, unresolved question rather than a marketing line.

The strongest independent number in the literature doesn't come from Bergen, or from the built-from-scratch Songdo, which wired vacuum pipes in from 2003 and has since fallen well short of its population targets. It comes from a modelling study the same 2023 review cites: an airport installation where an optimised compact terminal achieves up to 25% lower emissions than a conventional pneumatic terminal design. It took a peer-reviewed paper to surface that number. Nobody's marketing department got there first.

The catch

• Measured is optional, and most operators skip it. Rome, Prague and Bergen all report how many bins or properties they've connected. None reports what it saved. That gap isn't a rounding error, it is most of what a reader needs to know before believing any vendor's percentage.
• Drones catch dumpers; they don't handle waste. English councils have made illegal-dumping detection genuinely real: Maidstone's thermal-imaging drones logged 1,687 fly-tipping reports, 1,104 cleared within four days and 146 fines or prosecutions between April and September 2025, council-reported for the deployment period, not an independently audited before/after study. Wolverhampton's drone unit logged eight prosecutions in a year, and the national Environment Agency has raised its enforcement budget over 50% and secured a £1.4 million penalty against one serial dumper. Every other drone use in waste, marine-litter mapping, AI sorting, physical collection, is still research-stage or vendor pitch. Volvo and the waste company Renova announced a drone-plus-robot bin-emptying system with heavy press coverage in 2015–16; it never resurfaced afterward, no cancellation, no product, just silence, the pattern this atlas keeps finding in ambitious pilots.
• Pricing tools work precisely because they're visible, and visibility has a cost. Charging by weight or requiring a deposit changes behaviour because people notice the bill. That same visibility invites illegal dumping to dodge the fee and makes regressivity a live political fight that a quietly humming sensor bin never has to survive.

Verdict

On the evidence actually available, the cheapest tools in this article are also the best-proven. Germany's independently audited 96% bottle-return rate and Seoul's measured 25% cut in food waste are the two most solid numbers in the whole survey, and both run on pricing and a modest bit of hardware, not new pipes in the ground. Fill-level sensing is everywhere and probably works, the logic is hard to argue with, but the field has an evidence problem it hasn't solved: real deployments serving thousands of bins publish nothing, while the one rigorous academic trial that exists covers four bins on one street and a second rigorous trial found no benefit at all. Pneumatic tube networks are the genuinely extravagant bet, capital costs several times higher than trucks, an operating saving that's real but slow to repay it, and a flagship case in Bergen that, a decade in, still can't tell you what it bought. Drones have found exactly one job they're actually good at, catching people who dump illegally, and remain marketing everywhere else they're pitched.

Frequently asked