How AI is assisting demining operations in Ukraine

A large part of Ukrainian soil is now riddled with mines and shells. — AFP

After more than two years of conflict, Ukraine is now the most heavily mined country in the world, with almost a quarter of its territory contaminated by unexploded landmines. According to specialists, it would take more than 700 years to clear the country of mines using conventional methods. But artificial intelligence could help change the game.

Ukraine's first demining operations date back to the invasion of Crimea in 2014. Today, nearly 144,000km2 of land could be mined - almost a quarter of the country, according to NewScientist. To avoid the creation of semi-permanent exclusion zones, it is vital to find out exactly where to operate and which areas to prioritise for demining. That's why a team of researchers is developing the first AI model dedicated exclusively to mine clearance.

It takes into account a whole range of data, from army and emergency services records showing where bombs and shells have fallen, to agricultural maps and soil fertility data to identify the quality of the terrain to be cleared. The algorithm then assesses the importance and urgency of the land to be cleared, based on parameters such as population safety and potential economic gains.

Once key areas have been identified, demining specialists carry out more precise surveys of small plots, using drones equipped with ground-penetrating radar. At this stage, AI can only suggest priority areas for demining. But in times of urgent need, it can help save considerable time and energy.

One of the authorities' priorities is agricultural land. And unfortunately, the damage can be irreversible. Removing all traces of explosive ordnance is not always enough to compensate for the damage caused by bombing, starting with the large craters (up to several dozen meters deep) that ruin entire fields and even compromise the cultivation of surrounding land in the long term. – AFP Relaxnews