Intro

Hello and welcome to this introduction to Machines With Vision.

Like many other industries, digital transformation is sweeping through rail, offering passengers a more reliable and more efficient service. 

Geo-spatial data is critical to modern operation of railway infrastructure but current processes and data quality are outdated. 

We provide a geo-location platform that enables the digital transformation of railway infrastructure operations.

Do you recognise these asset mapping & condition monitoring challenges?

We have been working closely with experts at DB Netz in Germany and also Network Rail in the UK to understand the challenges of analysing infrastructure data recorded with on-train sensors.  

These are ... 

• Maintaining accurate maps of the railway assets

• Reliable and accurate geo-location of trains inspecting the infrastructure - these may be dedicated inspection trains or passenger trains equipped with sensors

• Repeatability of run-on-run position measurements 

• Operation at high railway speeds

• Low-cost sensor hardware

• High levels of automation and scalability of data processing

• Accuracy of the railway network model

• Guiding workers to specific track locations 

We’ve used this understanding to take our innovative geo-location technology and develop a geo-location platform enabling high performance solutions for infrastructure operators.

This is how our solution works

Each train is fitted with our camera sensor looking vertically down at the track. The cameras detect interesting features on the track called Visual Waypoints. A GNSS position sensor is also installed on the train. 

The camera and GNSS data is compressed and transferred over a mobile network to our cloud software where, starting from the first run, our algorithms calculate the track centreline and the geo-location of the Visual Waypoints with absolute accuracy improving run-on-run. 

For each individual train journey the train geo-location is calculated from the GNSS signal, corrected by our Visual Waypoints and recorded in a database.  

By time synchronizing condition monitoring and asset detecting sensors and accessing our track centreline and train movement data, high performance solutions for asset management, condition monitoring and maintenance are enabled.

The hardware

Here you can see our camera sensor which has been designed for installation on both measurement and passenger trains.

The highly robust and reliable sensor is fitted under the train looking vertically down at the track.

Case Studies

We have been trialling our solution with DB Netz in Germany and with Network Rail in the UK. 

In our UK trials we have benchmarked our system on thousands of kilometres of track and proven a 70mm run-on-run accuracy. This is 1 to 2 orders of magnitude better than a state-of-the-art satellite-based inertial navigation system used in the trial for comparison. 

We have also demonstrated accurate geo-location in tunnels, cuttings and in stations. 

Current trials have demonstrated operation up to 200 km/h and we will be increasing operational speed to 350 km/h, for high-speed train use, this year.

And in summary…

This is why we are the solution to your rail network asset mapping & condition monitoring needs:  

• Trainborne sensor measurements are geolocated, even in tunnels, stations and in adverse weather conditions with better than 1m accuracy

• Repeatability of geolocation is generally within 100mm accuracy allowing fully automatic alignment of run-on-run sensor data enabling, for example, predictive maintenance.

• We’ve proven the solution at 200 km/h and will be operating at 350 km/h this year

• Cloud based processing minimises train mounted hardware costs, 

• … is automatic and highly scalable allowing continuous monitoring even on large scale fleets of passenger trains,

• …furthermore from the first journey onwards generates and automatically updates an accurate up-to-date digital model of your railway network

• Additional functions, such as smartphone based applications, which allow track workers to locate themselves accurately along the track to verify a measurement or carry out maintenance, are additional capabilities we are developing

We have a technical solution and a flexible business model that scale to fulfil your requirements from small operational trials up to full network wide deployments.

And why Machines With Vision

Machines With Vision was founded in 2016 by Anthony Ashbrook, Tim Lukins, Jan Wessnitzer, and Matt Farrugia, of whom three hold PhDs in robotics, vision and artificial intelligence.  

The core team, which is 11 and growing, is truly international. We are building a company and a solution for a global railway market.