Landmines have been in use for hundreds of years as a military defense. In 1891, German chemist Carl Häussermann began experimenting with the explosive properties of Trinitrotoluene, more commonly known as TNT. Soon thereafter, German forces began weaponizing TNT, and the first TNT landmines were planted throughout Europe around the time of World War I. By World War II, tens of thousands of highly explosive landmines were laid across Europe. Landmines continued to be used in modern conflicts as well, despite efforts like the Ottawa Treaty, which saw 164 nations vow to ban the use of landmines in the late 1990s.

At the time of the Ottawa Treaty, it was estimated that approximately 160 million landmines, with little to no record of their placement, were hidden throughout the world. Meticulous and dangerous efforts have been carried out for decades to locate and remove these mines. Often, minefields are discovered only after a civilian steps on one while going about their day-to-day activities, triggering an explosion. Today, it is estimated that roughly 50 million landmines still exist worldwide. Due to the difficulty of locating these buried devices, researchers are constantly seeking novel methods for identifying and removing them.

One team of researchers from the University of Banja Luka in Bosnia and Herzegovina under the direction of Professor of Electrical Engineering, Vladimir Risojević, have begun using drone technology and bees to locate active landmines. Certain creatures, like rats and bees, can be trained to recognize the unique scent of TNT by associating it with a food reward. For Professor Risojević, the bees were trained to associate TNT with a sugar-water solution. Once released into a potential landmine field, the bees cluster around buried explosive devices in search of food, without triggering an explosion.

However, tracking where the bees cluster while keeping people safely out of harm’s way is not so easy. “We wanted to try to exclude humans from potential danger and try to use drones,” Professor Risojević said. To achieve this, he designed a multi-tiered experiment. The first step was to train a drone to recognize a small object, like a bee, flying through a field. To do this, Professor Risojević and his team first flew a drone across an open field in a grid pattern, capturing images to create a map. A duplicate map was made, onto which artificial images were overlaid to simulate bees. The team then developed an algorithm to train the computer to scan the data and track these simulated bees. The final step was to repeat the process, this time with live bees.

Professor Risojević published his findings in a paper titled A Method for Detection of Small Moving Objects in UAV Videos, in which he concludes: “The proposed method for detection of small moving objects in videos captured using UAVs opens up the possibility of its application to various honeybee surveillance tasks, such as pollination monitoring or landmine detection.” For the second phase of the experiment, drones were used to map a known, defused landmine while trained bees flew over it.

To create a controlled environment, a net was placed over designated areas at undisclosed test sites for the Croatian Mine Action Centre – Centre for Testing, Development, and Training. Within the large tent, healthy bee colonies trained to search for TNT were placed. Drones were then flown inside the tent along a grid several times. The footage was uploaded to the computer learning system. In his second paper on the experiment, titled Honeybee-Based Biohybrid System for Landmine Detection, Professor Risojević states: “The field experiments on test sites found that the spatial maps obtained using the active search show increased counts of honeybee detections at locations corresponding to the actual locations of the landmines.”

There are many methods for locating and defusing landmines, but combining drone technology with bees may not seem like the most obvious choice. Nevertheless, Professor Risojević recognizes its potential. “It’s very difficult for human observers to find these flying bees in video footage, let alone computer vision systems,” he explained. “There were moments when I thought we were outright crazy for trying to do that, but I am pleasantly surprised by the results we obtained.” The system proposed by Professor Risojević is not only unique but also presents a cost-effective way of keeping people safe when dealing with landmines.

Previous Drone News: