Australian railway

A major Australian railway operator wanted the ability to predict failures in the railway network a minimum of 24 hours in advance to reduce costs spent on preventative maintenance.
The operator approached us to create a minimum viable product (MVP) that took sensor data from turnout equipment to compare with failures.
A major Australian railway operator wanted the ability to predict failures in the railway network a minimum of 24 hours in advance to reduce costs spent on preventative maintenance.
The operator approached us to create a minimum viable product (MVP) that took sensor data from turnout equipment to compare with failures.

Results

• Automating pre-use checks reduces the likelihood of defects occurring
• Introducing machine learning means any defects that do occur are identified in less than one hour
• Automated tests replace time-intensive manual reviews, freeing up staff time for value add projects

British Board of Agrément (BBA)

BBA approached us to help bring together their siloed data after switching IT platforms. It needed a report that collated all the data that could be sent to everyone without needing to be manually created.
We created a new centralised system that allowed users across departments to get access to the information needed at the click of a button.
Discover how we transformed BBA’s siloed data

Results

Change Ahead

Change Ahead helps to connect vulnerable people with support services. To ensure they are showcasing relevant and high quality services, volunteers were spending hours manually searching for relevant websites, preventing the charity from scaling up.
Xerini proposed a scalable and accessible Al tool that acquires large amounts of useful data quickly and automatically validates relevant content.

Results

• Curating content each month used to require 40 volunteers -with Xerini’s platform it now takes three
• The one-to-one training calls for new volunteers that used to take up to an hour now take five minutes
• As the charity uses and develops the tool, the team will be able to auto-populate resources in all it’s regions, enabling Change Ahead to redirect hours of volunteer time towards new efforts

The Manufacturing Technology Centre

The MTC is an independent research and technology organisation that aims to bridge the gap between academia and industry. It approached Xerini for help improving machine set up and component quality on a metal 3D printing system.
Xerini created a system that verifies 3D printing baseplates using image processing techniques to make the set up more accurate and time efficient. We then developed a machine learning pipeline to detect possible defects, using a deep learning models to train the system. Xefr enabled us to integrate these systems into other existing ERP systems.

Results

• Automating pre-use checks reduces the likelihood of defects occurring
• Introducing machine learning means any defects that do occur are identified in less than one hour
• Automated tests replace time-intensive manual reviews, freeing up staff time for value add projects

Nationwide trade association

A 130-year-old trade association needed help with their digital transformation to increase the value of its services to members. They needed help providing a business case to its board of directors in order to invest in new digital technologies.
We provided guidance to define their business goals and how investment in digital transformation can help achieve those goals, including how they could best start. With our support, the material convinced the board to take a leap and invest in the transformation.
Find out how we offered our expertise…

Results

UK railway

A major UK railway operator wanted the ability to detect overgrown vegetation alongside rail tracks in order to schedule maintenance teams to cut them back. They approached us to create a minimum viable product (MVP) to process the images taken by a drone that flew along the track and work out which bits were vegetation and the extent to which they caused an obstruction.

We began making the LiDAR data available in the browser, rather than a desktop application so that stakeholders could have easy access to the images. We then ran further processing to detect and classify different aspects of the data, such as isolating the difference between the ground, buildings and foliage.
Discover more on how we displayed the LiDAR data…

Results