Artificial Intelligence in rail: innovative maintenance systems

Artificial Intelligence in rail: innovative maintenance systems

Artificial Intelligence is revolutionising rail maintenance systems in the UK. In this article we explore the latest technologies and innovations in the railway industry.

Up until recently, most site inspections carried out by Network Rail involved their workforce being on the tracks. This carries a significant level of risk for workers. Network Rail’s operatives receive intensive training before they can go on the tracks. Although this reduces the likelihood of accidents, the residual risk remains high. The main reason is the catastrophic nature of the consequences of an incident. As a result, Network Rail is promoting new technologies and innovations to minimise the time that site workers are exposed to risks.

Machine vision is set to change track maintenance

At the end of 2019, Balfour Beatty, in a joint venture with Omnicom and the University of York, unveiled what could be a game-changer for the rail industry. Together, they developed an Artificial Intelligence (AI) system that could save the industry up to £10M per year. This new system replaces traditional line inspection with a digitised method. Using machine vision, a camera attached to the front of a train captures high-quality images. An artificial intelligence system then processes. Finally, the system is able to highlight potential issues on the tracks.

Furthermore, the AI system can identify where problems may occur in the future. This brings the task of maintenance to a whole new level, where Network Rail’s team will have valuable insights to take preventive measures. An innovative solution like this can improve the accuracy and efficiency of site inspections. But, most importantly, it will reduce risks to rail workers by minimising the time they need to be exposed to high-risk scenarios on the tracks. Balfour Beatty’s project lead Stephen Tait said:

‘Our collaboration with the University of York has been invaluable; this latest innovation is an excellent example of how Balfour Beatty continues to deliver our commitment to reduce our onsite work by 25% by 2025 as we progress against our commitment to develop technologies to evolve the digital railway for a more reliable, cost efficient and safe network for all users.’

Richard Wilson, lead researcher on the project at University of York, also recognised that:

‘These machine vision technologies for high speed rail inspection will improve the reliability of the railway network, reduce costs and increase the safety of manual inspection.’

Artificial Intelligence and machine learning in rail

But the real ‘intelligence’ of the system does not stop at taking pictures of the rail tracks. Tim Flower, head of maintenance at Network Rail, said this year:

‘The key thing is to make sure we train those AI algorithms in the right way – making sure we’ve got the right insight in there.’

In fact, Network Rail is now going a step beyond with their Plain Line Pattern Recognition (PLPR), which uses not just cameras but lasers too, to film the railway. Rather than using machine vision alone to record what is out there in the field, they are also introducing machine learning. As they develop the artificial intelligence system, they have realised that it sometimes gives false-positives.

The machine learning means that the AI system will be able to remove false-positives. This way the final output consists of only real issues that need addressing by a human. Removing false-positives is also critical in another technological innovation that is gaining ground in the rail sector. We are referring, of course, to aerial surveys. As we know very well by now, drones are becoming very popular. Their LiDAR technologies are advancing quickly, and are now able to provide really accurate 3D surveys.

Drones and artificial intelligence

These aerial 3D surveys are very useful for multiple purposes. Network Rail seems particularly interested in using them to compare Digital Terrain Models (DTM) taken at different times. This helps them identify if the ground has moved, and how. The main advantage of this analysis is that it can identify and quantify potential geotechnical issues.

Another key benefit of the aerial 3D survey is that it can detect vegetation encroachment on the railway. Using more artificial intelligence on rail, maintainers can effortlessly identify the areas where there is vegetation encroachment. This, in turn, helps increase efficiency on the whole maintenance operation, by reducing the time traditionally spent on identifying these areas.

There is no doubt that artificial intelligence is revolutionising the rail sector. The industry is moving to a safer, more efficient and reliable position. Innovation and technology are at the forefront of these changes. What else will we see in the future? We cannot even imagine it now. But whatever that is, you will find it first here on our blog!

What other technological innovations have you come across recently? Let us know in the comments below!

Leave a Reply

Close Menu