Research ready to use
How can AI systems complement human performance?A user-friendly reference to freight train brakingSupporting the deployment of Assisted Braking and Door OperationQuick wins with Internet of Things sensorsMore holistic safety decisions by considering knock-on risk
Seven principles for effective human-AI teams in rail.
Artificial intelligence (AI) could transform rail operations. Potential applications include personalising customer service, optimising complex railway systems, and monitoring the condition of assets.
But to bring benefits, AI systems need to make the most of human strengths and mitigate human limitations. Introducing AI to rail without fully accounting for how it affects human roles could make railway jobs harder or less meaningful. Badly thought-through interactions between AI and humans can undermine their expected benefits and introduce new vulnerabilities. For example, it can leave staff ill-prepared for when situations demand human judgement.
We set out to identify a set of human factors principles for the design and operation of rail-specific AI systems. These principles aim to increase the likelihood that AI systems in rail optimise human performance and decision making.
Empower users and optimise their capabilities: AI should support, not sideline, human strengths.
Facilitate continuous people capability improvement: Users and AI can and should evolve together.
Create effective human-AI teams: Build AI as a teammate, not as a tool.
Set appropriate levels of explainability and transparency: AI should be able to tell you what it is thinking and why.
Build trust in the system: Trust in the system should be cultivated, calibrated, and maintained.
Define and overcome ethical challenges: Identify and design out bias, unfairness, and potential discrimination.
Build system and organisational readiness and manage change: Prepare the organisation and its people, not just the AI.
Story continues overleaf.
Find out more at https://rssb.co.uk/research-catalogue/CatalogueItem/T1362. You can download our introduction to the seven principles and read the full research report.
We are looking forward to refining the principles as they are applied in real-world settings. Please get in touch if you have feedback for us.
To discuss the principles and how to apply them in your own organisation, contact Paul Leach, Head of Human Factors:
Paul.Leach@rssb.co.uk
The knowledge gathered in one place will underpin improvements in safety and performance and inform future research.
Train braking is complex, involving a range of engineering systems, operational practices, train driver competencies, and wheel-rail interactions. The number of factors to consider also increases when trains run in low-adhesion conditions.
The Rail Accident Investigation Branch (RAIB) has investigated several serious GB freight train accidents in recent years related to braking and wheel slides, such as Petteril Bridge. Various workstreams are underway to provide a coordinated response to the RAIB’s recommendations. However, it was identified that there was only limited published information on freight train braking engineering and operational practices.
To fill this gap, we wrote an accessible reference guide to freight train braking. It provides an overall picture of freight train braking technologies, key terminology, driving approaches, and operational practices.
The work was carried out with the support of the Freight Braking and Adhesion Research Group as part of the cross-industry Wagon Condition Programme.
Download the freight train braking knowledge search at https://rssb.co.uk/research-catalogue/CatalogueItem/S386.
To discuss RSSB’s work on freight train braking, contact Paul Gray, Professional Lead, Engineering:
Paul.Gray@rssb.co.uk
This new Knowledge Search on freight train braking is full of detailed technical information but written in a manner which makes it approachable and understandable […] I have shared widely amongst my team here as essential reading for signal engineers—red lights don’t stop trains; the braking system does.
New guidance helps operators implement semi-autonomous systems more efficiently and safely.
Assisted Braking and Door Operation (ABDO) aims to provide some of the benefits of Automatic Train Operation without the infrastructure costs of full deployment.
Currently, stop car marker boards tell drivers where to halt the train along the platform. But different train types have different stopping positions, so there are often various signs along a single platform. The driver must remember the class of train they are driving, the train length, the location of the appropriate stop car markers on each platform, and the optimal braking point on their approach. Errors lead to stop short or overrun events, where the doors cannot be opened safely. And putting things right causes delays.
South Western Railway (SWR) and Alstom have been developing the ABDO concept for their Class 701 train fleet. The system detects when the train is approaching a station, calculates the braking force required to stop the train in the correct position at the platform, and applies the brakes accordingly. The driver always retains control of the train and can apply further braking force or override the ABDO if necessary. The ABDO system automatically opens the doors when all conditions are met.
Other operators are also considering ABDO as part of new train procurement. To gather the understanding needed for wider adoption of ABDO, we explored the impact of its introduction on performance, capacity, and safety. This work included reviewing the Rule Book and other relevant standards. Although we recommend some updates to explicitly cover the new method of operation, we found nothing that would prevent ABDO’s introduction to the network.
ABDO has the potential to improve performance by reducing dwell times. ABDO functionality is also expected to reduce risk at the platform-train interface associated with overrunning and stop short incidents.
Using the research findings and the experience of SWR, we have developed a good practice guide. It is designed to help operators anticipate and prepare for the steps needed in ABDO implementation. It provides structured guidance through the implementation journey, from initial planning to operational deployment and beyond.
Download the good practice guide from https://rssb.co.uk/research-catalogue/CatalogueItem/T1326.
To discuss the research, contact Marcus Carmichael, Professional Lead, Operations and Performance:
Marcus.Carmichael@rssb.co.uk
Our new Class 701 fleet supports the ABDO system, but its impact on the driving task and wider network operations must be understood. This research explored these effects across various scenarios to help guide our future decisions.
Removing barriers to fitting inexpensive sensors that can improve efficiency and passenger comfort.
Internet of Things (IoT) sensors are inexpensive and can equip older rolling stock with modern capabilities. The sensors collect and share real-time data, such as temperature, humidity, power, noise, and vibrations. They can enable predictive maintenance, enhanced operational efficiency, and improved passenger comfort.
However, there are barriers to IoT sensor adoption. These include difficulty navigating standards, identifying who to involve, and mounting sensors onto rolling stock.
We produced guidance to help engineers overcome these challenges. The guidance covers all stages of the IoT sensor implementation process, from early identification of goals through to maintenance and lifecycle management.
After the type of IoT sensor is identified, the reader is guided through a potential procurement process. This is followed by compliance and installation guidelines, including technical considerations, mounting methods, and the engineering change process.
Next, the guidance covers data management and cybersecurity. Finally, the reader is guided through good practice maintenance, lifecycle management, and continuous monitoring.
The guidance aims to breaks down barriers to IoT sensor implementation, helping more operators and rolling stock owners benefit from IoT sensors.
Download the guidance at https://rssb.co.uk/research-catalogue/CatalogueItem/T1330.
RSSB is looking at trialling the guidance to test its effectiveness and usefulness. A trial would also be an opportunity to explore the integration of data from sensors. If you are interested in working with us on a trial, please get in touch. Contact Tom Preece, Research Analyst:
Tom.Preece@rssb.co.uk
A new study has improved the quantification of secondary risks arising from delays.
Measures to control the immediate safety risk from situations that could harm passengers, staff, or the public often cause delays. For example, if someone falls from the platform edge, all trains due to arrive at that platform need to be held or diverted until the accident has been dealt with. But as delay filters through the system, that can itself result in ‘knock-on’ risk, for example, by making stations more crowded or presenting more red aspects to drivers.
This research improves the quantification of knock-on risk. This supports decisions about how to manage a range of delay-causing events. The research looked at both the short- and long-term effects of delays on safety.
For short-term risk, we compared safety incident rates with delays. Days with more delays had higher rates of safety incidents. We quantified this relationship using a regression model and 3 years of data. It was possible to calculate:
the potential reduction in safety incidents if delays were eliminated
the growth rate of safety incidents for every additional minute of delay.
Long-term effects were assessed using data on how rail demand changes in the face of delays and disruption. By combining this with the fatality risk of other transport modes, we quantified the risk from passengers choosing alternative, less safe modes of transport.
The research findings help make balanced operational decisions that consider different ways in which safety is impacted. The outputs can also support wider business cases and government policy decisions.
The final report can be downloaded from https://rssb.co.uk/research-catalogue/CatalogueItem/T1344.
To discuss the implications of this research on your own operation, get in touch with Liz Davies, Professional Lead, Data and Modelling:
Liz.Davies@rssb.co.uk
I believe that every operator should read and understand this report. This research provides a factual evidence base to quantify the safety risks that delays can import. It will help us take balanced risk decisions that consider the different ways our actions can affect safety.
