Interview of Prof. Francesco Flammini published on Railway Gazette about the Europe's Rail project "Roadmaps for A.I. Integration in the Rail Sector" (RAILS) that he led as the Technical Manager.
1. 46 November 2023 Railway Gazette International
IN FOCUS Technology
Integrating AIinto
therailsector
A European research project led by Italy’s National
Interuniversity Consortium for Informatics suggests that
adoption of AI technologies will have a transformative impact
on railway business and operations, as project manager
Francesco Flammini explains to Murray Hughes.
Intelligent
maintenance was
the first main area
of investigation for
the researchers,
who hope that
by implementing
predictive
maintenance
strategies driven by
AI, operators can
achieve a substantial
reduction in
unplanned downtime.
I
n a world driven by technological
advancements, the railway industry
is embracing the power of artificial
intelligence to transform operations,
enhance safety measures, and optimise
performance.
The Rails (Roadmaps for AI
Integration in the Rail Sector) research
project, which concluded in June,
explored how AI technologies could
be adopted in the rail sector and what
the impact would be.
Scope and objectives
Rails chose a multifaceted approach to
tackle the core challenges faced by the
rail industry, utilising cutting-edge AI
technologies such as machine learning
and deep learning. The project set out
to achieve three primary objectives:
intelligent maintenance, enhanced
safety measures, and optimised
operations.
The project was carried out under
the Horizon 2020 Framework
Programme, specifically within the
Shift2Rail Joint Undertaking call for
proposals. Launched in December
2019, it brought together a consortium
of four research partners in Italy, the
UK, the Netherlands and Sweden. An
advisory board included research
institutions, railway operators,
technology companies and
organisations specialising in data
analytics and AI integration.
The project was co-ordinated by
CINI, the National Interuniversity
Consortium for Informatics,
represented by Prof Valeria Vittorini,
and technically managed by Prof
Francesco Flammini.
Rails received funding from the
European Union’s Horizon 2020
research and innovation programme
under grant agreement No 881782. Its
objective was to promote the
development and integration of
cutting-edge AI technologies in the rail
sector, aligning with the European
Commission’s vision for sustainable,
safe, and efficient transport.
Main research areas
Intelligent maintenance was the
first main area of investigation. By
implementing predictive maintenance
strategies driven by AI, operators
can achieve a substantial reduction
in unplanned downtime, leading
to cost savings and enhanced asset
performance. Early fault detection
enables proactive maintenance
interventions, minimising disruption
and ensuring smoother operations.
The field of intelligent train control
covered co-operative train driving and
artificial vision, both of which play a
pivotal role in improving efficiency and
detecting potential safety hazards and
security threats. Real-time monitoring
and anomaly detection through AI can
provide immediate alerts to operators,
enabling timely responses and the
application of preventive measures.
Advanced monitoring and surveillance
capabilities have the potential to
contribute to develop autonomous trains
in open environments, considering all
relevant threats and uncertainties.
Investigation of optimised
operations found that the integration
of AI-powered optimisation algorithms
had the potential to revolutionise rail
operations and traffic management.
Intelligent scheduling and routing
systems yield improved punctuality,
reduced congestion, and better
resource allocation. In practice, this
means passengers can experience
smoother journeys, while freight
Photo:
ProRail
/
Jos
van
Zetten
2. Railway Gazette International November 2023 47
Technology IN FOCUS
operators can benefit from enhanced
logistics efficiency.
Flammini believes that ‘the impact of
the Rails project extends beyond the
consortium partners. Its comprehensive
roadmaps, guidelines, and best practices
provide valuable guidance for rail
operators, policymakers, and stakeholders
across Europe. The dissemination of
knowledge and insights derived from
the project have aimed to foster a wider
adoption of AI integration in the rail
sector, leading to a more intelligent and
sustainable transport network.’
Intelligent maintenance
By harnessing the power of AI algorithms
and data-driven approaches, the project
aimed to facilitate predictive maintenance
strategies, leading to lower operational
costs and better asset reliability. Rails
investigated advanced data analytics
techniques, such as machine learning
algorithms, to analyse a wealth of sensor
data, historical records, and external
factors like weather conditions.
These sophisticated AI models
accurately predict potential equipment
failures, enabling proactive maintenance
interventions that minimise downtime
and optimise asset performance. The
integration of AI in maintenance practices
is instrumental in achieving significant
cost savings, extending asset lifespans, and
enhancing overall operational efficiency.
Flammini points out that ‘the research
focus of Rails in intelligent maintenance
encompassed various aspects, including
developing predictive maintenance
models, optimising maintenance
scheduling, and integrating sensor data
analysis techniques. The research
outcomes shed light on the potential of
AI in transforming maintenance
practices. For instance, the use cases
focused on the development of predictive
maintenance models that incorporate
historical data, sensor readings, and
machine learning algorithms to
accurately forecast potential equipment
failures. This allows railway operators
to implement proactive maintenance
strategies, resulting in reduced downtime
and improved asset reliability.
‘Another research effort explored the
optimisation of maintenance scheduling
using AI-driven algorithms, taking into
account factors such as asset condition,
operational requirements, and resource
availability. These studies demonstrated
the efficacy of AI in supporting
maintenance practices and providing
practical insights that can be applied by
rail operators.’
Intelligent train control
Safety has always been paramount in
the rail industry, and Rails recognised
the potential of AI integration to
improve safety measures. The project
devoted substantial research efforts to
address advanced AI-powered train
control systems. By leveraging computer
vision algorithms, machine learning
techniques, and real-time data analysis,
Rails explored the opportunities offered
by AI to empower operators with the
ability to monitor tracks, infrastructure
and rolling stock in real time.
AI-driven systems can facilitate swift
identification of safety hazards and
security threats, helping to ensure safer
journeys for passengers and safeguarding
critical infrastructure. They can also
significantly augment safety standards,
providing operators with real-time
insights to address potential safety
concerns promptly and prevent incidents.
‘The research focus of Rails in
enhancing safety measures through AI
integration explored various topics,
including co-operative train driving
through virtual coupling and artificial
vision for obstacle detection’, says
Flammini. ‘The research outcomes of
Rails demonstrated representative pilot
use cases showing the potential efficacy
and limitations of these approaches when
applied in the rail sector. For example,
one research effort focused on developing
an automated anomaly detection system
that utilised AI algorithms to analyse
video footage and identify potential
obstacles. This can enhance safety
measures and enable prompt responses
to potential threats in open environments
where there are many uncertainties.
‘Other opportunities relate to the
development of intelligent risk
assessment models that incorporate
real-time data from multiple sources’,
continues Flammini. ‘These models
provide operators with a holistic view
of potential safety risks and allow for
proactive mitigation measures. We
believe the research findings contribute
to the development of AI-based safety
solutions and provide rail industry and
operators with valuable guidance on
defining realistic targets.’
Optimised operations
Rails placed great emphasis on
optimising operational efficiency
within the rail sector through AI
integration. To achieve this, the project
Fig 1. A wide range
of railway-related
problems was
investigated in the
Rails project.
Fig 2. An intelligent
maintenance system
for level crossings
was one area of
investigation.
3. IN FOCUS Technology
48 November 2023 Railway Gazette International
developed intelligent scheduling
and routing algorithms, leveraging
predictive analytics and optimisation
techniques. By analysing historical
data, demand patterns, disruption
and infrastructure constraints, Rails
enabled rail operators to make data-
informed decisions on scheduling,
routing, and resource allocation.
AI-driven optimisation models can
offer insight into what action needs to be
taken, resulting in reduced congestion,
enhanced punctuality, and better overall
operational effectiveness and, ultimately,
better service for passengers.
Flammini explains that ‘the research
focus of Rails in optimising rail
operations through integration of AI
explored topics such as demand
forecasting, network optimisation, and
real-time decision-making algorithms’.
He notes that ‘the research outcomes
offer valuable insights into the potential
benefits of AI-driven optimisation
strategies in the rail sector’, citing the
development of a demand forecasting
model that accurately predicts
passenger volumes.
‘This allows operators to allocate
resources efficiently and optimise
service frequency, leading to improved
use of capacity and enhanced customer
satisfaction’, he continues. ‘Rail network
optimisation algorithms can minimise
travel times, reduce congestion, manage
disruption effectively and improve
connectivity between different routes.
Research findings provide rail operators
with practical approaches to optimise
operations and deliver enhanced service
experience to passengers.’
Collaborative approach
Rails thrived on a collaborative
approach, fostering partnerships between
industry experts, researchers, railway
operators, and technology companies.
This collective effort was instrumental
in ensuring an effective analysis of the
state-of-the-art, railway industry needs,
relevant use cases, and future roadmaps.
The project also served as a platform
for stakeholders to pool their expertise
and perspectives, resulting in a realistic
view of the AI potential and limitations
in practical applications.
‘The collaborative approach embraced
by Rails has not only advanced the
integration of AI in the rail sector but has
also set the stage for future collaboration
between academia and industry,
fostering ongoing innovation and
driving the adoption of transformative
technologies’, Flammini says.
He believes that the collaboration
fostered by the Rails project has
yielded valuable insights and outcomes
that extend beyond the individual
research papers. The exchange of
knowledge, best practice and lessons
learned among stakeholders has
contributed to a broader understanding
of the challenges and opportunities
associated with AI integration in the
rail sector.
‘The collaborative approach has also
facilitated the transfer of technology
and expertise, ensuring that AI-based
solutions are practical and aligned with
the operational realities of the rail
industry’, suggests Flammini.
Positive outlook
Rails has far-reaching implications for
the rail sector and paves the way for a
future where AI integration becomes
the norm, the research team believes.
By demonstrating the effectiveness
of AI in intelligent maintenance,
enhanced safety measures, and
optimised operations, it has unlocked
new possibilities for the rail industry.
The research outcomes, published
in various papers, provide a solid
foundation for future initiatives,
guiding policymakers, researchers, and
industry professionals in their pursuit
of AI-driven innovations.
‘Moving forward, the rail sector must
embrace the findings and lessons from
the Rails project to drive further
advancements in AI integration’, asserts
Flammini. ‘The successful
implementation of AI-powered
solutions has the potential to transform
the rail industry into a safer, more
efficient, and customer-centric mode of
transport. Rail operators should
consider the practical implications and
feasibility of adopting AI technologies
within their existing infrastructure,
while policymakers and industry
associations can play a pivotal role in
creating a conducive regulatory and
funding environment to support AI
integration in the rail sector.’
Future role
Flammini believes that ‘ongoing research
and development efforts should focus on
areas such as dataset sharing, cognitive
digital twins, mixed-reality testing and
trustworthy AI technologies in the
rail industry. Collaboration between
academia, industry, and government
institutions will continue to play a crucial
role in advancing the state-of-the-art in AI
integration and addressing the challenges
associated with its implementation.’
In addition, ‘by leveraging the
momentum created by the Rails project
and building upon its key research
focus, the rail sector can embark on a
transformative journey towards a future
where AI integration is fully realised,
benefitting operators, passengers, and
the entire transport ecosystem’.
Flammini is convinced of the huge
potential offered by AI in next-
generation railways, pointing out that ‘as
the rail sector embraces the insights and
lessons learned from the Rails project, it
is poised to unlock the full potential of AI
technologies. By incorporating intelligent
maintenance practices, enhanced safety
measures, and optimised operations, rail
operators can significantly improve their
efficiency, reduce costs, and enhance the
overall passenger experience.
‘Ongoing research and development,
coupled with collaborative initiatives, will
ensure that the rail industry can continue
to evolve and leverage AI integration.
Rails has laid a solid foundation for this
transformation, and the future holds
immense possibilities for the rail sector
as it harnesses the power of artificial
intelligence’, he concludes.
Fig 3. AI-based
architecture for
‘self-protection
mechanisms’ at level
crossings.
primary
objectives
were set
for the Rails
project:
intelligent
maintenance,
enhanced
safety
measures
and optimised
operations
3