Alstom Cuts 5,000 Wildlife Collisions in Sweden with AI

STOCKHOLM, SWEDEN – Alstom and Swedish startup Flox Intelligence are conducting advanced field tests of an AI-powered wildlife detection and deterrent system in Sweden. As of April 2026, the project has entered a new implementation phase on the Dalabanan and Bergslagsbanan lines, deploying both AI-assisted video cameras and tailored acoustic warning signals. The initiative, supported by operator VR and transport authority Tåg i Bergslagen, targets a reduction in the approximately 5,000 wildlife-train collisions reported annually across the country.

What Are the Technical Specifications?

The system combines real-time video analysis with a responsive acoustic deterrent to clear animals from railway lines. The core technology uses AI-powered cameras to identify animals near the tracks. Once an animal is detected and classified—such as a moose, deer, or wild boar—the system activates specific audio signals designed to scare that particular species away. The machine learning algorithms are continuously trained, improving accuracy with each identification, and have already demonstrated high performance in identifying smaller animals and birds. The total project cost and the effective range of the detection and acoustic systems have not been disclosed.

Key Technical Data

Where Does This Technology Stand in the Market?

The Alstom-Flox AI-camera system enters a market where sensor-based and simpler acoustic systems are more established. Competing technologies often rely on different detection methods. For example, the UK-based Pro-Tect system uses passive infrared (PIR) sensors to trigger acoustic alarms, a proven but less specific method. In contrast, the Flox system’s primary advantage is its AI-driven visual identification, allowing it to distinguish between species and potentially ignore non-threatening objects, reducing false positives. Other market solutions include RADAR-based detection systems, which are effective in all weather but can be more costly to deploy over long distances. The “tailored audio” component of the Flox system represents a step beyond the generic alarm sounds used by many existing systems. (Source: Manufacturer Data, 2024).

Editor’s Analysis

This project reflects a significant shift from passive mitigation measures, like fencing, to active, intelligent deterrent systems in railway safety. The focus on machine learning highlights a broader industry trend of leveraging AI to solve persistent operational problems, moving beyond just predictive maintenance into real-time hazard prevention. While Sweden’s 5,000 annual collisions are a key driver, this technology has clear export potential to other regions with high wildlife-rail interaction, such as North America and other parts of Europe. (Source: European Union Agency for Railways Safety Report, 2023).

FAQ

Q: What specific animals is the AI system being trained to detect?
A: The system is being trained on a wide range of species found in Sweden, including large animals like moose, deer, and wild boars. During initial tests, it also proved highly accurate in identifying smaller animals like foxes, as well as domestic animals and various bird species.

Q: How does the system scare animals away without harming them?
A: The system uses tailored acoustic signals, which are specific sounds designed to be unpleasant or alarming to particular animal species, encouraging them to leave the area. This method avoids physical contact and is intended as a humane deterrent to prevent collisions.

Q: Will this technology be deployed across all of Sweden’s railway lines?
A: A nationwide deployment plan has not been officially confirmed. The current tests are focused on specific lines with high accident rates, and the results will likely determine the feasibility and scope of a broader rollout.