Latest / Internet of Things with Fexingo: Connected Devices, Sensors, and Industrial IoT / How IoT Sensors Are Detecting Potholes Before They Form
Transcript
- Lucas: So there's this pilot project in Helsinki that I think is genuinely one of the smartest uses of IoT I've seen in a while. They're embedding sensors directly into the road surface to predict potholes — not just find them after they form, but actually catch the conditions that lead to them. Luna: Wait — they can predict a pothole before it happens? How does that work mechanically? Lucas: It's a combination of two sensor types. First, accelerometers that measure micro-vibrations in the pavement — essentially the road's 'strain' as vehicles pass over it. When the readings start to show a pattern of increased flexing, that's a sign the underlying base layer is weakening. Second, moisture sensors that track how much water is seeping into the pavement layers. Luna: And moisture is the real trigger with freeze-thaw cycles, right? Lucas: Exactly. In Helsinki, they get a lot of winter freeze-thaw — water gets into a crack, freezes, expands, then thaws, and that repeated cycle is what turns a small crack into a crater. The sensors detect when moisture levels are high AND temperatures are oscillating around freezing, and the system flags those road segments for preventive sealing before the pothole ever opens up. Luna: So it's less about spotting a pothole and more about spotting the recipe for a pothole. I love that. What kind of numbers are we talking — how many sensors, how much area? Lucas: The pilot covered about 50 lane-kilometers of major roads and used roughly 2,000 sensor nodes. They're about the size of a deck of cards, embedded about 10 centimeters below the surface during resurfacing work. The city reports that in the first full year — that's 2025 — they reduced pothole formation by 30 percent compared to adjacent control segments that didn't have sensors. Luna: That's a pretty dramatic drop for a pilot. And I'm guessing the cost savings go beyond just not having to fill holes — fewer lawsuits from damaged tires or suspensions, less traffic disruption... Lucas: Helsinki estimated they saved about $2.7 million in direct repair costs last year alone. And that doesn't include the indirect stuff like reduced vehicle damage claims or lower emissions from crews not having to drive out to patch roads. The preventive sealing treatment costs about a tenth of what a full pothole repair costs. Luna: That's compelling — but I have to ask, for smaller cities with tighter budgets, is this even feasible? Those sensors and the data infrastructure can't be cheap. Lucas: You're right to ask, and the cost has been the biggest barrier. But some of the design work is going open source. There's a project called OpenRoad-IoT that publishes sensor schematics and data processing pipelines for free. The hardware cost has dropped to about $15 per node if you're building them yourself. Helsinki's pilot used commercial units that were more like $80 each, but the open-source approach is bringing that down fast. Luna: And if a city can't afford to instrument every road, they can at least put sensors on the trouble spots that have a history of potholes — kind of a targeted approach. Lucas: That's exactly what some cities in the US are doing now. Pittsburgh started a small deployment last year on streets that had more than three repairs in a single winter. Their hope is to move from reactive patching to a maintenance schedule based on actual data, not just calendar intervals. And this is where the machine learning layer comes in. Luna: Right — you're collecting all this vibration and moisture data, but how do you turn that into an alert that a crew can act on? Lucas: The sensors feed into a model that's trained on historical pothole occurrence data. It learns patterns like: if the accelerometer shows a certain frequency deviation AND moisture exceeds X percent AND temperature has dropped below freezing twice in the past 24 hours, then there's an 80 percent chance of a pothole forming in the next week. The system then generates a work order for the preventive sealing crew. Luna: So it's almost like a weather forecast for pavement. I wonder if this kind of data could eventually be shared with navigation apps — like Waze or Google Maps — to warn drivers about high-risk sections even before any damage is visible. Lucas: That's actually being discussed in Helsinki. The city has an open data portal, and they're exploring partnerships with mapping companies to add a 'pothole risk index' layer. But there's also a longer-term possibility that autonomous vehicles could use this data to adjust their route or speed when approaching a known high-risk zone. If a self-driving car knows a road is weakened, it could slow down to reduce the dynamic load. Luna: That's fascinating — the infrastructure and the vehicles start to communicate. But let's bring it back to the here and now. What's the biggest hurdle to scaling this beyond pilots? Lucas: I think it's the data integration piece. Right now, most city public works departments don't have the software to ingest real-time sensor data and turn it into actionable maintenance schedules. You need someone — or some system — to look at that data every morning. Helsinki hired a data analyst just for this pilot. For a city like, say, Buffalo, New York, that's a new line item. Luna: And I could see the maintenance crews also needing training to trust the sensor alerts over their own experience. There's always a cultural shift when you introduce data-driven decisions. Lucas: Absolutely. One of the early challenges in Helsinki was that the road crews were skeptical because the sensors would flag a section that looked perfectly fine. But when they did a core sample, they found micro-cracking starting below the surface. After a few of those cases, the trust grew. Now they rely on the dashboard as their primary planning tool. Luna: That's a great proof point. And it reminds me of something that often comes up on this show — the real value of IoT isn't just the sensor itself, but the system around it that changes how people work. Lucas: Exactly. And that's actually something that I think ties into why we're able to keep this show going without ads. The depth we get to explore — like the Helsinki project — comes from listener support. If today's tech conversation gave you something usable, you can help keep these episodes coming at buy me a coffee dot com slash fexingo. It's a simple way to support the kind of deep-dive reporting we do here. Luna: Yeah, it truly makes a difference. Every contribution goes straight into research time and booking guests. Lucas: So back to the road data — one more stat I want to share. Helsinki is now planning to expand the sensor network to cover all major arterial roads by 2028, and they're even experimenting with solar-powered wireless nodes that don't need to be wired into the grid. That would make deployment in rural areas much cheaper. Luna: And theoretically, if those nodes can communicate over long-range low-power networks like LoRaWAN, you could monitor highways between cities without running fiber. That could be a game-changer for state departments of transportation. Lucas: Right. The traditional approach is to send an inspector out once a month to visually assess road condition. With IoT, you get continuous data at a fraction of the labor cost. The Department of Transportation in Minnesota is already testing a similar system for frost heave monitoring on rural routes. Luna: Frost heave is a huge problem in the northern states — it can cause serious road damage and even accidents. So the same sensor stack could address multiple failure modes. Lucas: Exactly. And the really intriguing thing is that once you have this sensor infrastructure in place, you can piggyback other use cases on it. Helsinki is already adding air quality sensors to some of the road nodes, and they're testing pedestrian counting modules for crosswalk optimization. Luna: So the road network becomes a platform, not just a piece of asphalt. That's the kind of infrastructure thinking that I think more cities need to adopt. Lucas: Yeah, and it makes the upfront investment more justifiable when you can spread the cost across multiple services. A single sensor node that does pothole prediction, air quality, pedestrian counting, and maybe even traffic flow — that's a much easier sell to a city council than a single-purpose device. Luna: Especially when you can point to a 30 percent reduction in potholes and $2.7 million in savings. Those are real numbers that translate to taxpayer benefit. Lucas: So the question that sticks with me is: if this works in Helsinki, what's stopping every major city from adopting it within the next five years? Is it just a matter of political will and budget cycles, or are there technical barriers we haven't solved yet? Luna: I think the technical barriers are mostly solved in the pilot phase. The real challenge is organizational — getting different departments to share data, training crews, and standardizing the sensor formats so that cities aren't locked into a single vendor. But as more open-source designs become available, that lock-in risk decreases. Lucas: And there's also the question of who pays for the initial deployment. Helsinki's pilot was funded partly by the city and partly by a European Union smart-city grant. For a cash-strapped city, the payback period might be two to three years, which is reasonable, but the upfront capital is still a hurdle. Luna: That's where public-private partnerships could come in — maybe a contractor installs the sensors in return for a share of the maintenance savings over time. I've seen similar models in energy efficiency projects. Lucas: Interesting model. And it gets even more interesting when you factor in autonomous vehicles, because those vehicles need good road conditions to operate safely. If AV companies want to expand their service areas, they might have a financial incentive to help cities deploy this kind of monitoring. Luna: So the pothole sensor might end up being a piece of the autonomous vehicle infrastructure puzzle. That's a pretty big leap from filling holes in the road. Lucas: It is. And it's exactly the kind of cross-domain connection that makes IoT so exciting. The sensor data that starts as a pothole predictor becomes a road health index, which becomes an input for self-driving navigation, which becomes a dataset for city planning. Each layer adds value without needing new hardware. Luna: I'm sold. Let's hope more cities start paying attention to what Helsinki is doing. For now, I'll be looking at my local potholes a little differently. Lucas: Same here. Every time I hit one, I'll think about the accelerometer data we're missing.