Imagine a train yard where hundreds of freight wagons need to be sorted and assembled into new trains — it's like a giant puzzle done mostly by hand, often at night, and it's slow. SMART built a camera-and-laser system that can spot obstacles on the tracks up to 1000 meters away, even in the dark, and recognize individual wagons with centimeter precision. On top of that, they created a web-based control room system that plans the whole sorting process in real time, so yard operators finally get a digital dashboard instead of clipboards and radio calls.
Rail freight marshalling yards are bottlenecks — sorting and assembling trains is slow, labor-intensive, and often unsafe, especially during night operations. Yard operators lack real-time digital tools to plan and optimize marshalling, and there is no reliable automated system to detect obstacles or recognize wagons during shunting. This drives up transport time, increases costs, and limits the competitiveness of rail freight against road transport.
The project built two main things: a multi-sensor obstacle detection prototype combining thermal cameras, image intensifiers, stereo vision, and laser scanners for detection up to 1000 m; and a web-based demonstrator for marshalling yard supervision and management, complete with database, visual yard representation, traffic data input, process planning, and integration interfaces for railway IT systems.
If you are a rail freight operator struggling with slow turnaround times at marshalling yards — this project developed a real-time yard management system that optimizes resource planning and marshalling operations. The web-based platform supports TAF/TSI standard data formats, meaning it connects to your existing network systems and allows shared usage of yards between different service providers.
If you are a rail infrastructure company dealing with collision risks during shunting and low-speed operations — this project built a sensor fusion prototype combining thermal cameras, image intensifiers, stereo vision, and laser scanners for obstacle detection up to 1000 meters, working day and night. The system also provides wagon recognition at short range with +/- 5 cm distance estimation tolerance, directly applicable to automated shunting.
If you are a sensor technology company looking for proven multi-sensor fusion architectures — this project combined four different sensor types into one obstacle detection system covering mid-range up to 200 m and long-range up to 1000 m. The algorithms handle impaired visibility, night operations, and precise distance measurement at +/- 5 cm, which is transferable to mining vehicles, port equipment, or other heavy machinery.
The project data does not include pricing or cost-per-unit figures. The system was developed as a prototype under Shift2Rail funding. Any commercial deployment would require negotiation with the consortium partners, particularly Universität Bremen as coordinator.
The project delivered a working demonstrator of the web-based yard management system, including database storage, visual representation of yards, traffic data input, and integration interfaces for railway company IT systems. Scaling to a full yard would require engineering work beyond the prototype stage, but the architecture was designed with real operational requirements in mind.
The project was funded under the Shift2Rail programme (Shift2Rail-RIA). IP rights are governed by the consortium agreement between the 5 partners. Licensing inquiries should be directed to the coordinator, Universität Bremen, or the relevant partner holding specific IP.
Based on the project objective, the sensor fusion system combines thermal camera, image intensifier, multi-stereo vision, and laser scanner specifically to handle day and night operation as well as impaired visibility conditions. Long-range detection reaches up to 1000 m and mid-range up to 200 m.
The system supports TAF/TSI standard data formats for connection to external network systems. It was designed to enable shared usage of marshalling yards between different service providers, which means it was built with interoperability as a core requirement.
The project delivered prototype-level solutions: a sensor fusion prototype for obstacle detection and a web-based demonstrator for yard management. These are functional proof-of-concept systems, not commercial products. Further development and certification would be needed before operational deployment.
The SMART consortium is compact — 5 partners across Germany, Bulgaria, and Serbia, heavily weighted toward academia with 4 universities and just 1 industry partner (20% industry ratio). This is a research-driven project, which means strong technical depth but limited commercial pull. The coordinator, Universität Bremen, is a well-established German research university. The inclusion of 1 SME suggests some commercial awareness, but a business looking to adopt this technology should expect to bring its own commercialization resources and expect further development beyond what the project delivered.
Universität Bremen, Germany — reachable through CORDIS or the project website
Want to connect with the SMART team about licensing their obstacle detection or yard management technology? SciTransfer can arrange a direct introduction and help structure the conversation.
