Safety Where and When It Matters Most
Trespassing on railway tracks, whether intentional or accidental, presents a significant safety risk. When people enter the track area, rapid detection and immediate action are essential to prevent potentially life-threatening situations.
To address this challenge, ZÖLLNER is working with an AI-powered obstacle detection solution based on compact edge computing technology. The system runs on NVIDIA Jetson Nano hardware and uses a real-time computer vision model from the YOLO family to continuously monitor the surrounding area and identify people in the track zone within fractions of a second.
Keeping People Safe Is Our Mission
Detection is only the first step. Once a person has been identified, the system can assess the situation and initiate appropriate safety measures. The specific response depends on the operational environment and the level of risk involved.
Possible actions include:
- Triggering audible or visual warning signals
- Sending alerts to train drivers or control centers
- Interfacing with railway signaling systems
- Initiating operational interventions, up to and including emergency braking
By combining artificial intelligence, real-time image analysis, and automated response mechanisms, obstacle detection creates new opportunities to identify hazards at an early stage and react before incidents occur.
Let’s Work Together and Save Lives
You provide the problem and we supply the safety solution!
If you know of any situation where trespass detection could help keep people safe, please take a minute and fill out the form below.
Where We Started
As part of the REAKT initiative in Schleswig-Holstein, ZÖLLNER partnered with the Christian-Albrechts-University of Kiel to explore the potential of artificial intelligence for future railway applications. The project focused on developing a reliable, cost-efficient train detection system that could support intelligent railway infrastructure and autonomous rail operations.
Combining affordable sensors, camera technology, and AI-powered image analysis, the research team evaluated more than 400,000 images captured under a wide range of weather and lighting conditions. Through continuous optimization of the training data, the system achieved a train detection rate of 85% at 30 frames per second, while processing images directly without the need to store video recordings.
The result is a smart and resource-efficient solution for reliable train detection, bringing the railway industry one step closer to intelligent, automated infrastructure. The project demonstrates how machine learning and artificial intelligence can help enable safer, more efficient, and more economical railway operations in the future.
Read more about ZÖLLNER and the REAKT Project here
Safety Is Our Priority!
Get in touch today so we can talk about how to keep people safe tomorrow!
Fon: +49 431 7027-137
E-Mail: [email protected]
This article was originally published by ZÖLLNER.
