Location and data positioning in the railway

In railway operations accuracy is paramount, especially when tracking trains along their routes for maintenance tasks and defect detection. Whether it's ensuring the right track is reported or pinpointing a fault along the line, precision monitoring is the backbone of safe and efficient railway management.

How does the railway use location?

From signalling to maintenance, every aspect relies on precise positioning. These include:

1.     Signalling: Ensures signallers (and other trains) know where the trains are for performance and collision avoidance

2.     Timetabling / passenger information / delays: Helps in scheduling trains and informing passengers about arrivals and departures

3.     Infrastructure monitoring: Allows monitoring of railway tracks and structures for safety and maintenance purposes

4.     Maintenance work: Guides maintenance crews to specific locations requiring attention

Each of these tasks require different levels of accuracy, but at the very least, we need to know which track the train is on. The railway operates like a one-dimensional line - a train can't move left or right off the track.

If the wrong track is mistakenly identified, the consequences can range from data loss through to speed restrictions and worse. Sometimes, it's okay not to measure every inch, but we need to be aware when the position is wrong or uncertain, otherwise we may make decisions based on incorrect data. Pinpointing faults is therefore essential for effective maintenance and safety, as even a marginal error in location can lead to significant problems.

Systems in place

Nowadays, position invariably is provided by GNSS (Global Navigation Satellite Systems) including GPS, Glonass, and Galileo. These satellite systems identify where trains are, providing highly precise positional accuracy – potentially down to a few centimetres. But embankments, buildings, tunnels, OLE and other parts of infrastructure can degrade or block the signal, making it hard to obtain accurate measurements.

Systems like ETCS, Driver Advisory Systems (DAS) and Automatic Train Operation (ATO) utilise a combination of GNSS, wheel encoders and balises (meticulously placed along the track) to provide highly accurate reference points within a metre or so and with the correct track identified. These systems rely on physical infrastructure as well as additional on-vehicle equipment.

When it comes to Infrastructure Monitoring, we need to know exactly where a fault is so that we can either a) monitor it or b) fix it. To do either of these, we need to understand the accuracy of the data and if there is any ambiguity in the positional information. Manual validation is typically required - ultimately leading to data loss when the data is unreliable or ambiguous.

Sources of error

Despite advancements, errors in GPS and other positioning systems can cause issues. For example, with GNSS/INS measurements, errors in position aren't the only factors affecting accuracy. Simply having latitude and longitude coordinates might not be very helpful without additional context. These coordinates need to be matched to a network model or interpreted by a device like a handheld GNSS tool for maintenance tasks.

Furthermore, the accuracy of GNSS is influenced by various factors such as satellite visibility, environmental conditions like ionospheric events, the type and maintenance of equipment and even the time of day. If a maintenance worker is given coordinates to fix a fault, but they are inaccurate, finding the fault becomes difficult. Factors like built up urban areas and tunnels make it even trickier.

Map matching tools can help and aim to find the best fit for the input data against a network model, but with errors in both data sets, they don’t always produce the right answer.

When measuring potential defects using Infrastructure Monitoring, it’s worth noting that a seemingly small synchronisation error (between the measurement and positioning systems) can lead to significant positional inaccuracies, especially when trains are moving quickly. For instance, at 125mph, a 10-millisecond error translates to nearly 56cm of error. Accurate timestamping of data from input sensors is essential to maintain accuracy in Infrastructure Monitoring.

With all these potential errors, certainty of position and track is not guaranteed, it raises the question: How can we accurately determine the track and position of the train or fault?

Correction services and train sensors

Various methods can enhance railway positioning accuracy. Correction services like RTK and PPK reduce GNSS errors, but they're not always available. Physical infrastructure like balises and RFID tags provide location data but require installation and maintenance.

On-train sensors like LiDAR can determine a train's direction through switches, but their effectiveness may vary in complex infrastructure, and they also rely on an accurate network model. Lineside fibre optic cables can also be utilised for positioning but require installation and calibration.

Ideally, we need a universal solution that minimises physical installations while maintaining accuracy everywhere in near real-time – especially when data needs to be updated to match other infrastructure changes.

Introducing RailLoc

RailLoc revolutionises railway positioning by leveraging computer vision technology to interpret track features such as ballast structures and bolt orientations. By identifying such elements, it can determine the precise location of the train along the track.

This method offers absolute accuracy without the need for physical interventions. It’s universally applicable, eliminates manual processes, detects infrastructure changes like tamping effects and track renewals, operates in near real-time and ensures precise positioning even in GNSS-denied areas like tunnels.

RailLoc’s accompanying Fault Navigator app closes the loop on fault identification even in challenging environments. Fault Navigator is a pocket version of RailLoc and enhances worker safety by directly and precisely navigating them to faults – whether located in a straightforward open sky environment, or more challenging location such as a tunnel or urban area with multiple tracks. With RailLoc and Fault Navigator, railway maintenance becomes more efficient and accurate, promising significant improvements in safety and reliability.

To find out more about how our systems work in action, please get in touch with Jon Owen, CTO at Machines With Vision, via jon.owen@machineswithvision.com or by connecting directly on LinkedIn.