CZ/SK verze

Sweden Turns to AI to Prevent Thousands of Rail Wildlife Collisions

Sweden Turns to AI to Prevent Thousands of Rail Wildlife Collisions
photo: ALSTOM/Photo: ALSTOM/ALSTOM technicians installing equipment
20 / 05 / 2026

Sweden’s railways record around 5,000 collisions with wild animals every year, causing delays, material damage, and psychological strain for train drivers. A new artificial intelligence system developed by Alstom and Flox Intelligence aims to prevent these incidents before they happen.

AI Patrol on the Front of the Train

Railway lines crossing Nordic landscapes often become deadly traps for wildlife. In Sweden alone, approximately 5,000 collisions between trains and animals occur every year.

This persistent problem is now being addressed with artificial intelligence. Alstom has partnered with Swedish startup Flox Intelligence to deploy a system of smart cameras capable of actively preventing tragedies on railway tracks, according to Alstom.

The technology is based on advanced machine vision. Cameras mounted on moving trains continuously scan the surroundings of the railway line in real time. As soon as the system detects a living creature near the tracks, it immediately launches an acoustic warning response.

Unlike conventional deterrent systems, which animals eventually learn to ignore, the new technology relies on surprise and behavioural psychology. The AI identifies the specific animal species and selects the most effective warning signal for it. The sound waves constantly change frequency to prevent animals from becoming accustomed to them. The system uses authentic sounds of natural predators, instantly triggering escape instincts in wildlife.

Moose and Deer Presented the Biggest Challenge

Although the system performed well in recognising birds and domestic animals during testing, wild Nordic fauna proved more difficult for developers. The algorithm required considerable training to correctly identify deer and moose. However, thanks to continuous data collection, the digital system is constantly improving and becoming more accurate with every kilometre travelled.

"We were surprised by how many animals we were able to observe and record from a single train set," said Maria Signal Martebo, Managing Director of Alstom Sweden.

The collected data is also revealing how effective—or ineffective—existing physical fencing along railway lines actually is.

Fewer Delays and Reduced Psychological Pressure

The project, financially supported by Swedish innovation agency Vinnova, entered its second phase in April 2026. Live testing is currently taking place in cooperation with operators Tåg i Bergslagen and VR on four major Swedish railway lines. The economic and human benefits of the technology could be substantial, reducing financial losses, operational delays, and psychological stress. For train drivers, collisions with large animals are often deeply traumatic experiences.

Tags