Latest / Internet of Things with Fexingo: Connected Devices, Sensors, and Industrial IoT / How IoT Sensors Are Detecting Land Mines
Transcript
- Lucas: You know how we talk a lot on this show about IoT sensors making factories safer, farms smarter, supply chains more efficient. Luna: Right — kind of the quiet productivity revolution. Lucas: Exactly. But there's a use case that flips that script entirely. Instead of preventing a leak or optimizing a process, these same sensors are being used to find buried land mines. Luna: That's a dramatic shift — from industrial safety to literal life and death. Lucas: Yeah. And it's not theoretical. A non-profit called the HALO Trust, which is the oldest humanitarian demining organization in the world, has been piloting a sensor network in Ukraine's Kharkiv region. They're using off-the-shelf vibration sensors and metal detectors — the kind you'd find in a factory for predictive maintenance — networked together with mesh radio. Luna: So they're not even building custom hardware. They're adapting stuff that already exists. Lucas: Exactly. And that's the whole point. The sensors are cheap — we're talking maybe 50 dollars per node. They're deployed by drone, dropped in a grid pattern over a suspected minefield. Each node has a small solar panel and a battery. They communicate with each other and relay data back to a base station up to a kilometer away. Luna: So the deminers don't have to walk into the field to check the readings. They stay at a safe distance. Lucas: Right. The traditional method is terrifying — someone in a protective suit uses a handheld metal detector and prods the ground with a stick. It's incredibly slow and dangerous. With this sensor network, they get a map of vibration anomalies and metallic signatures across the whole field, and they can prioritize which areas to clear first. Luna: But land mines are small. How sensitive do these sensors have to be? Lucas: That's the hard part. The vibration sensors can pick up footsteps — human or animal — but they also pick up wind, rain, vehicle rumble from nearby roads. The metal detectors trigger on anything metallic: shrapnel, spent cartridges, old farming equipment. The false positive rate is enormous. Luna: So how do you separate a real mine from a rusty nail? Lucas: That's where the sensor fusion comes in. They use machine learning models trained on thousands of hours of vibration data from known mines and non-mine objects. The system learns to recognize the specific signature of a pressure plate being disturbed versus a piece of metal shifting in the wind. It's not perfect, but it reduces the number of false positives by about 70 percent compared to a single sensor. Luna: And that 70 percent — that means deminers spend much less time digging up harmless scrap. Lucas: Exactly. Time is the most scarce resource in demining. The HALO Trust estimates that with traditional methods, clearing a single square meter of minefield costs about 10 dollars and takes a full day. With the sensor network, they can cover the same area in a few hours and reduce the cost by half. Luna: That's a huge efficiency gain, especially in a country like Ukraine where the scale is staggering. I read that since 2022, Ukraine has become the most mined country on earth — more than Afghanistan or Syria. Lucas: Yeah, it's a nightmare. Ukraine's economy ministry estimates that about 30 percent of the country's land is potentially contaminated with mines or unexploded ordnance. That's roughly 174,000 square kilometers. At current demining rates, it would take over 700 years to clear it all. Luna: So technology like this isn't just a nice to have — it's the only way the timeline becomes manageable. Lucas: Precisely. And the same sensor mesh concept can be used for other things. Once the mines are cleared, the same network can be left in place to monitor soil moisture, temperature, even detect unauthorized digging — which is a problem for theft of archaeological artifacts or illegal logging. Luna: That's smart — the infrastructure pays for itself beyond the initial mission. Lucas: And that's a theme we often see in industrial IoT: the same sensors can do double duty. But I think the most impressive part is that this is already working in the field. In the Kharkiv pilot, which ran from September to December last year, the sensor network helped clear 12 acres of agricultural land. That land is now being used to grow wheat again. Luna: That's a concrete outcome you can point to. Not a proof of concept — actual farmland returned to use. Lucas: Yeah. And that pilot was done with just 200 sensor nodes. The HALO Trust is now raising funds to scale it to 5,000 nodes across multiple regions. The cost per node is low enough that even a modest donation can equip a whole field. Luna: Which brings up something — if today's conversation has given you a new angle on how sensor tech can be used, or if you appreciate that we keep this show ad-free and focused on real-world engineering, listener support is what makes that possible. People who find value here can toss a few bucks at buy me a coffee dot com slash fexingo. It genuinely helps us avoid running ads. Lucas: And it lets us spend time on stories like this — the ones that don't get the same venture capital attention as smart refrigerators but arguably have a bigger impact. Anyway, back to the tech — one of the cleverest parts of the system is how they handle battery life. Luna: Because in a minefield, you can't exactly send someone out to swap batteries. Lucas: Right. So each node has a small solar panel — about the size of a paperback book — and a supercapacitor instead of a lithium-ion battery. Supercapacitors charge faster, last for hundreds of thousands of cycles, and are less sensitive to extreme temperatures. They're not as energy-dense, but for a low-power sensor that only sends data every five minutes, it's plenty. Luna: And supercapacitors are also cheaper and safer — no fire risk if a node gets crushed. Lucas: Exactly. The whole system is designed to be robust, cheap, and disposable if necessary. Each node costs about 45 dollars in parts. Compare that to a traditional mine detection robot, which can run 50,000 dollars or more. You can deploy a thousand nodes for the price of one robot. Luna: And the robot still has to be driven by a human at a safe distance, so it's not truly autonomous. The sensor network is more like a smart carpet that you spread over the danger zone. Lucas: Yeah, that's a good analogy. And the data from the carpet can be visualized as a heat map — red areas where the sensors detected something suspicious, green where it's likely clear. Over time, as the machine learning model sees more data, the maps get better. The HALO Trust has already released a dataset of over 10,000 sensor readings to help other researchers improve the models. Luna: So it's open-source in spirit, even if the hardware is commercial off-the-shelf. Lucas: Exactly. And that openness is key because the problem is global. There are an estimated 110 million land mines buried in 60 countries. Most of them are in places that can't afford expensive custom solutions. If you can build a demining sensor network from parts available on AliExpress, that's a game changer. Luna: What about the data transmission? If you're in a remote field in, say, Angola or Cambodia, you might not have cellular coverage. Lucas: That's why they use mesh networking — specifically a protocol called LoRa, which stands for Long Range. It can transmit data up to 15 kilometers line of sight on very low power. Each node acts as a repeater, so the signal hops from sensor to sensor until it reaches the base station. No cell towers needed. Luna: So the network is self-forming and self-healing. If one node dies, the others route around it. Lucas: Exactly. And that's another parallel to industrial IoT — same mesh topology used in smart factories to avoid a single point of failure. But in this case, the failure point is literally land mine that destroyed a node. Luna: Which is a vivid reminder of why we need to clear these things in the first place. A farmer or a child steps on a mine, and it's catastrophic. The sensors are basically taking that risk for them. Lucas: Yeah. And that's the heart of it. We spend a lot of time on this show talking about optimization — saving 5 percent on energy, reducing downtime by 10 percent. But sometimes the optimization is about saving a leg, or a life. And the same technology does both. Luna: That's a good note to end on. Thanks, Lucas. Lucas: Thanks, Luna. See you next time.