AI models in Trimble Business Center software were trained to extract 20,000 rail sleepers from a point cloud, drastically reducing the time needed to identify anomalies. Rail lines are engineered to evenly distribute the massive weight of locomotives and cars carrying tons of freight and passengers every day. Small variations in track width or height can lead to inefficient energy use and vibrations, while more serious issues may result in derailments and high-speed accidents.
To improve safety and assist with longterm lifecycle management, innovative technology companies like Rhomberg Sersa Rail Group (RSRG) are adopting new methods of monitoring, inspecting and analysing rail infrastructure. They are leveraging cutting-edge technology, including 3D laser scanners and artificial intelligence (AI), to conduct these critical tasks both during and after construction.
Gotthard Base Tunnel project
In August 2023, Switzerland’s Gotthard Base Tunnel sustained significant damage during a 16-car derailment that shut down all passenger and cargo travel through the western tube for over one year. The 35-milelong rail tunnel – the longest in the world – is an important transportation link between Germany in the north and Italy in the south.
RSRG is an international expert in end-to-end rail construction, providing a range of services including reality data capture, support planning, construction and deformation monitoring. After the accident at Gotthard, the firm was contracted to complete the repair work, including dismantling the damaged track and installing and aligning a new Low Vibration Track (LVT system) over a four mile distance. All contractors were under pressure to complete construction quickly so tunnel operations could resume. The high profile project required high accuracy and careful logistical planning under challenging conditions with high humidity and poor air quality inside the tunnel.
During the entire renovation of the Gotthard Base Tunnel (GBT), rigorous testing and control measures were implemented to ensure the safety of the restored infrastructure. This included inspecting all 20,000 sleepers and verifying that the spacing between each pair met the 60-cm tolerance requirement (+/- 2 cm).
Dimitrios Kyritsis, R&D product owner in RSRG’s Digital Rail Services department, explained his company’s philosophy: ‘Our Digital Rail Services department embraces technological advancements to make operations smarter, more efficient, and safer as travel speeds increase. We focus on leveraging technology to meet customer needs while optimising workforce deployment to save costs and reduce risks.’
After considering the options, RSRG felt manual measurements would require too much time and would delay the tunnel’s opening. Instead, the surveyors opted for a more efficient method. They chose to scan the tunnel using a Trimble GEDO IMS-Scan track measurement system. Then, they leveraged a new capability within Trimble Business Center (TBC) software to train custom point cloud classification models to automatically extract the sleepers from the point cloud.
The track measurement trolley was equipped with a Trimble GEDO GX50 laser scanner, which uses dual scan heads to produce an exceptionally dense, 360-degree point cloud. To ensure complete coverage, each scan head was set at an 80-degree orientation, and the tunnel was scanned twice – once in each direction. The Trimble GEDO system also includes an Inertial Measurement Unit (IMU), which captures highly accurate positioning data with 3–5 mm accuracy while scanning.
The dual-profile scanner system is ideal for as-is and as-built reality capture. It allows for detailed comparison with design models and quality checks of a wide range of features, including concrete surfaces, power lines, rails, sleepers, rail fasteners and rail inclination.
While RSRG had previously used the Trimble GEDO Track and Trimble GEDO IMS systems to produce final as-built documentation with +/- 1 mm accuracy, this was their first time training the new AI feature extraction tool in TBC.
Automate workflows with custom AI training
The Trimble GEDO track measurement system produced 900 point cloud files (500 GB of data) to be processed in Trimble GEDO Scan Office. With an Excel file for control, individual georeferenced point clouds were merged into a large data set. To help isolate the sleepers, noise was removed from the point cloud, retaining only the ground features around the track.
Manual feature extraction would have been extremely time-intensive and prone to errors due to the sheer volume of data. Instead, RSRG opted to use a new tool in TBC that allows users to train 3D deep learning models and customise feature extraction workflows for specific needs.
This new tool leverages AI to mimic human cognitive functions, with the goal of producing results that are as high-quality as those from human intelligence. These significant productivity gains are accessible to everyone without requiring AI expertise.
To automate the measuring process, RSRG created sample point cloud data containing manually classified sleepers to train the AI model in TBC. They applied the model to the complete point cloud collected with the Trimble GEDO system to automatically classify the sleepers that had been installed in the tunnel.
‘Experimenting with and testing the new tool was quite easy. Even without prior AI knowledge or background, it is straightforward to learn by following numerous free online webinars and supporting documents’ said Kyritsis. ‘A statistical analysis assessed the training process and indicated whether we needed a bigger area and more training. It was important to us to be able to verify the accuracy of the results, so it wouldn’t be a ‘black box’ for the engineers. We were pleased reaching a 97 per cent accuracy after training the model.’
After applying the AI model to extract the sleepers, RSRG’s in-house-developed modules compared the 3D model against the reality capture data. Using a best fitting approach, these modules calculated measurements based on the absolute coordinates from the Trimble GEDO track measurements.
The result was a comprehensive, 340 page report that calculated distances at a mm/sub-cm level for all pairs of sleepers. It specifically highlighted those that did not meet the 60-cm tolerance requirement (+/- 2 cm).
‘Other software limits the kinds of features we can classify and extract’ said Kyritsis. ‘TBC offered the capability to train AI for a specific type of sleeper and improve the results. AI allowed us to automate the extraction process and complete the work quickly with confidence in the results.’
AI adds value to Big Data
There is growing demand for smarter ways to handle large volumes of data as adoption of 3D laser scanners and mobile mappers continues to increase. Automation using AI is a great time saver when processing and analysing high density reality capture data.
The benefits of using AI for feature extraction in TBC extended far beyond the Gotthard Base Tunnel project. Since the trained models can be easily distributed to other offices, RSRG created value for the entire company.
With these training files readily accessible, inspections can now be performed more frequently, allowing teams to be proactive rather than reactive in their maintenance efforts.
Fabrice Lardon, digital rail expert at RSRG, is sure about the positive contribution of AI to the railway industry, stating that ‘AI-powered tools, such as the new feature extraction capabilities in TBC, have the potential to revolutionise railway industry inspections, providing engineers with advanced solutions to elevate their work to the next level.’
The Trimble GEDO track measurement system captured data quickly, helping to avoid further delays and enabling the tunnel to reopen on schedule. This comprehensive dataset now serves as a valuable record of current conditions, which can be compared with future data collections to guide proactive maintenance.
The western tube of the Gotthard Base Tunnel reopened in September 2024 and is now able to accommodate 260 freight trains and 70 passenger trains per day. Moving forward, digital reality capture data will play a critical role in making informed decisions at every stage of the tunnel’s lifecycle management.
