Innovative Drone Program Combats Global Landmine Crisis with AI Precision

Safe Pro launches program to use drones, AI in landmine detection

By DRONELIFE Features Editor Jim Magill

Landmines, which kill or injure as many as 100,000 people in nations across the world every year, represent a global health crisis.

Experts at the organization behind the international Minesweepers estimate that currently there are 110 million land mines in the ground in about 60 countries. Locating and removing mines by traditional methods involving humans equipped with metal detectors and a big dose of courage, is dangerous, expensive and highly inefficient.

A start-up drone and artificial intelligence (AI) company, using off-the-shelf drones and machine learning algorithms is striving to rise to the challenge of introducing a technology-based solution to the worldwide landmine crisis.

“We’re eager to share the work that we’re doing and the results of what we’re doing and make a significant impact in the world,” Dan Erdberg, CEO of Safe Pro Group, said in an interview. Safe Pro’s subsidiary, Safe Pro AI, is currently conducting a drone-enabled demining pilot program in war-torn Ukraine.

Using off-the-shelf small aerial vehicles and its propriety technology, SpotlightAi, initial results have proven to be promising, locating 60 or more mines in one of its initial field tests in the Ukrainian countryside, said Jasper Baur, Safe Pro AI’s co-founder and lead scientist.

 SpotlightAi harnesses machine-learning models to process massive amounts of sub-centimeter-level captured by commercially available drones. Powered by the Amazon Web Services cloud, the platform is able to instant processes drone-collected images, using “an extensive proprietary dataset of hundreds of thousands of labeled images of over 150 various types of landmines and unexploded ordnance,” according to a recent company press release.

The types of UXO that can be identified by the system range from small, anti-personnel devices known as butterfly mines – devices that when trigged can fly upwards and explode, killing or maiming any unfortunate farmer or child who happens to come into contact with them – to American-supplied unexploded cluster munitions.

The SpotlightAi technology instantly detects, labels and assigns a GPS tag to any surface mines and other explosive remnants of war and generates detailed reports that provide actionable intelligence to the appropriate government entities and non-governmental organizations engaged in demining operations.

Safe Pro officials made a presentation on the development of on SpotlightAi at a recent meeting of the Geneva International Centre for Humanitarian Demining. The prestigious group acknowledged that the technology represented a breakthrough in the dangerous and import business of detecting landmines, Baur said.

“Right now, no one’s using any kind of AI for analyzing their drone imagery,” he said. “The benefit of SpotlightAi is that it provides the analysis almost instantly,”

In order to develop the machine-learning model, Safe Pro AI’s team labeled more than 100,000 different items and made images showing the items from different viewpoints to teach the software how to distinguish a landmine or other piece of unexploded ordinance from other objects found in real-world environments. “It creates this general idea of what a grenade is, what a rock is, what an anti-tank mine is,” Baur said. With these models in its electronic mind, the software uploads imagery captured with a drone, applies that learned information to identify all potential hazards on a map.

The system is built for use with any commercial off-the-shelf aerial vehicles. The SpotlightAi team has employed the system with drones from a variety of providers, including DJI and Parrot. One of the chief benefits of the Spotlight Ai system is that it’s compatible with essentially any drone equipped with a visible-light camera with 12-megapixal or greater resolution, Erdberg said.

Ukraine

In war-torn Ukraine, Safe Pro teams have been working for the past several months with the nation’s Ministry of Defense and The HALO Trust, a non-governmental organization dedicated to international demining efforts, to prove the efficacy of the SpotlightAi system. The scope of the problem in that country is huge. Experts estimate that large sections of the country, equal in area to the state of Florida, are seeded with hidden unexploded ordinance, likely to pose a deadly danger for a long time, even after the hostilities eventually end.

Experts estimate that it could take up to 800 years to completely clear the country of unexploded ordinance using traditional methods.

“The World Bank has allocated over a billion and a half dollars for demining in Ukraine,” Erdberg said. “There’s a lot of activity right now going into how do we tackle this problem efficiently and effectively. And they’re looking at new technologies.”

Bauer, in his fourth year of a PhD program at Columbia University, has had a longtime interest in the nascent field of landmine detection as a future career path, although his primary field of doctoral research involves volcanos.

“I studied demining first,” he said. “And then, since my background was in geology, I went into graduate school in volcanology because there’s no program for landmine detection. It’s just very uncommon.”

As it turned out, the study of volcanos proved to be valuable in developing his anti-landmine system, as both mines and volcanos represent potential explosive hazards that endangered humans. “I developed a lot of skills that I used doing studying volcanology that I use every day for detecting mines and doing the mining work,” he said.

His early fundamental research proved the feasibility of using drones equipped with visual and thermal imagery detect equipment to detect foreign objects lying on the surface of the ground. The challenge then was how to scale up the drone-based mine-detection model into a system that could be duplicated on a large scale.

“How is this something that can really reach millions of people and be a useful skill set? That is where the machine learning comes in,” Baur said. With friend and colleague, Gabriel Steinberg, Baur cofounded Safe Pro AI to develop a machine learning model that can automate the detections of landmines and other unexploded ordinance.

Safe Pro Group

Safe Pro AI is one of three affiliated companies in the Safe Pro Group, the other two being SafeProUSA and Airborne Response. SafeProUSA builds body armor and equipment used in explosive ordnance disposal and mine-clearing. Airborne Response is a Florida-based drone services provider that provides critical mission services for customers who don’t want to assemble their own fleet of drones.

Erdberg said the three companies work in tandem to provide a complete end-to-end solution to the vital job of dealing with unexploded ordinance. “We can provide you the drone services, we can provide you the AI and ML, and then we can provide you the ballistic protection you need to eventually go and bring the human into the loop to clean the minefield,” he said.

Jim Magill is a Houston-based writer with almost a quarter-century of experience covering technical and economic developments in the oil and gas industry. After retiring in December 2019 as a senior editor with S&P Global Platts, Jim began writing about emerging technologies, such as artificial intelligence, robots and drones, and the ways in which they’re contributing to our society. In addition to DroneLife, Jim is a contributor to Forbes.com and his work has appeared in the Houston Chronicle, U.S. News & World Report, and Unmanned Systems, a publication of the Association for Unmanned Vehicle Systems International.

Read more: