A type of land mine called the “butterfly” has a particularly insidious reputation for two reasons: It is known for killing or crippling children who may pick up what looks a lot like a green plastic toy, and its mostly nonmetallic construction means it often evades traditional mine detectors. Butterfly mines’ light-touch detonators go off easily if stepped on by a fighter — or farmer — and their relatively small charge often maims people without immediately killing them.
More than a million Russian-made PFM-1 land mines — the most common butterfly type, possibly inspired by similar U.S. weapons deployed during the Vietnam War — still litter Afghanistan after decades of conflict. During the Soviet-Afghan War in the 1980s, military helicopters dropped swarms of these mines, whose “wings” let them flutter to the ground. This created instant minefields blocking high mountain passes, contributing to the problem of land mines and unexploded ordnance responsible for killing or injuring more than 30,000 Afghan civilians since 1978. In recent years children have made up the majority of victims killed or wounded by such weapons in Afghanistan.
Land mine clearance usually involves a careful, meticulous, time-consuming process of interviewing local residents and then sweeping suspected areas twice on foot, using handheld mine detectors. The estimated costs of removing a single mine can range from $300 to $1,000 — and even confirming a patch of ground is safe without finding any mines costs money. But now, early tests by a group of U.S. researchers have shown they may be able to make the process a lot faster and more efficient by using a thermal imaging camera mounted on a quadcopter drone.
“The only way to clear mines is to poke every inch of the ground,” says Alex Nikulin, assistant professor of energy geophysics at Binghamton University in New York. “But we can tell you where to poke.”
Nikulin and his team are basing their strategy on the fact that plastic heats up and cools down differently than the surrounding soil and other parts of the environment. By conducting flight tests around sunrise and sunset, when temperatures can rapidly change, the researchers showed that inert PFM-1 land mines lying on the ground become clearly visible in thermal imagery captured by the drone camera and displayed on a laptop. They presented their work at the annual meeting of the American Geophysical Union, held in Washington, D.C., in December.
In field trials conducted in September 2017 at a New York state park, the team was able to pick up almost 78 percent of the mines during four trials. That is not yet good enough to fully replace survey work by ground teams. But it could help narrow down the locations and general layout of PFM-1 minefields, says Alex van Roy, deputy head of operations at the Geneva-based Swiss Foundation for Mine Action (FSD) — and a mine action specialist who formerly served in the Australian Army.
“This drone with this tech may allow you to determine in a much more rapid fashion where the minefield footprint is,” says van Roy, who was not involved in the Binghamton team’s research. “You find the perimeters of the footprint so that you can focus your manual, expensive clearance team in as small an area as possible.”
The idea of using thermal imagery to detect land mines has been explored before but has only recently become more practical, as both drones and thermal imaging cameras get smaller and less expensive. A next step could use computer software to automatically detect land mine signatures in the thermal imagery, Nikulin says. His team hopes to recruit coders to develop a machine-learning algorithm capable of handling image recognition patterns. The goal is to eventually offer a $10,000 boxed set — which would include the off-the-shelf drone, infrared camera and a laptop loaded with custom software — to humanitarian demining organizations.
Cost matters a lot. Some of FSD’s demining work in Afghanistan and Iraq gets U.S. State Department funding, but the Swiss foundation still relies heavily on donors — and on spending its funding wisely. “If [the technology] costs $50,000, we can’t use it,” van Roy says. “That’s enough to purchase 10 traditional land mine detectors, or maybe fund a team of humans in Afghanistan for a month.”
There are other complicating factors. Some plastic land mines originally dropped decades ago may have been buried in landslides or earthquakes, van Roy says. Sunlight exposure may have degraded others to the point where their shapes defy easy visual identification. Governments also often restrict drone use in conflict zones within countries such as Afghanistan or Iraq.
Still, the Binghamton team’s focus on Afghanistan is a good strategy, van Roy says. Thermal imaging from the air may not work for buried land mines or cluttered environments with a lot of vegetation or human-made artifacts — but it does seem better suited for spotting the PFM-1 land mines dropped in Afghanistan’s dry, high-elevation mountain passes with their generally sparse population and relatively light underbrush.