ERTMS/ATO is Europe’s interoperable solution for automatic train operation, but it depends on ERTMS/ETCS to provide the train protection and signalling information required by ATO. This creates a migration issue: many lines are still operated with national signalling systems and lineside signals, while ERTMS/ETCS trackside deployment takes time.

Lineside Signalling Interpretation is one possible migration concept to bridge this gap. It does not replace ERTMS/ATO, nor does it create a new target architecture; it aims to reuse the standard ERTMS/ATO stack by feeding an ERTMS/ETCS onboard unit with information derived from existing lineside signals.

This article explains why this concept should be understood less as a perception technology and more as a migration architecture towards interoperable railway automation.

Recommended reading

For a good understanding of the concepts discussed in this article, I recommend reading first:

1. ERTMS: the European Rail Traffic Management System
2. ERTMS/ETCS: the European Train Control System
3. ERTMS/ATO: Europe’s Interoperable Train Autopilot

Executive Summary

ERTMS/ATO is designed as the European interoperable solution for Automatic Train Operation. In the target architecture, ATO Onboard receives the information it needs through the ERTMS/ETCS onboard system, while ETCS supervises the train inside its safe movement envelope. The principle remains clear: ATO drives, ETCS protects.

This architecture is coherent for the future European railway system. But it creates a migration challenge. ERTMS/ATO depends on the availability of ERTMS/ETCS, while many lines remain operated with national signalling systems and lineside signals. Full ERTMS/ETCS trackside deployment is necessary, but it is expensive, slow and will take many years.

Lineside Signalling Interpretation is a migration concept designed to reduce this dependency. The idea is to equip the train with an ERTMS/ETCS onboard unit and to feed it with signalling information derived from existing lineside signals. A perception component detects the signal aspect. A signalling converter translates this information into ERTMS/ETCS-compatible data, using train localisation and infrastructure data from a Digital Map. The ERTMS/ETCS onboard unit can then provide the information required by the standard ERTMS/ATO onboard function.

From an architecture perspective, this can be understood as a train-centric emulation of ETCS Level 1. In conventional ETCS Level 1, the signal aspect is interpreted trackside by a Lineside Electronic Unit, and the corresponding ETCS telegram is transmitted to the train through Eurobalises. In Lineside Signalling Interpretation, part of this interpretation chain is moved onboard, using perception, localisation, Digital Map data and conversion logic.

The value of this concept is not to create a new proprietary automation stack. It is to preserve the ERTMS/ATO target architecture while introducing a migration layer between existing lineside signalling and the ERTMS/ETCS onboard unit. This could support earlier use of automatic driving, strengthen the business case for onboard ERTMS/ETCS equipment and prepare some of the capabilities needed for future automation.

However, Lineside Signalling Interpretation is not the final target system. It cannot provide the same capacity performance as radio-based ERTMS/ETCS. It is limited to areas where lineside signals can be detected and interpreted. It requires high-quality infrastructure data, safe localisation, robust perception, clear operational rules, hazard analysis, certification principles and careful management of degraded situations.

Its purpose is therefore not to replace ERTMS/ETCS trackside deployment. Its purpose is to support the transition. Lineside Signalling Interpretation is a migration bridge between today’s lineside signalling and tomorrow’s interoperable automated railway system.

Last modified: 2026-06

Content

1. Why the migration problem exists
2. Lineside Signalling Interpretation as a migration concept
3. A train-centric emulation of ETCS Level 1
4. The system architecture
5. From signal aspect to ETCS-compatible data
6. Migration benefits
7. Limits and open points
8. Architecture perspective
9. Conclusion
10. Documentation and further reading

1. Why the migration problem exists

ERTMS/ATO was designed as the European interoperable solution for automatic train operation. Its purpose is to avoid the creation of national or supplier-specific automatic driving systems on mainline railways. Instead of building one automation solution per network, per signalling system or per operator, the European approach is to define a standard ATO architecture working together with ERTMS/ETCS.

This is essential because mainline railways are open systems. They involve multiple railway undertakings, infrastructure managers, fleets, signalling systems, traffic management environments, operational rules and migration constraints. A metro line can often be automated as an integrated system. A mainline railway cannot.

The target architecture is therefore clear enough in principle. ERTMS/ATO automates the driving task. ERTMS/ETCS provides the train protection layer. The driver remains in the cab in GoA2, supervising the operation and taking over when needed.

But this target architecture depends on ERTMS/ETCS.

On lines equipped with ERTMS/ETCS trackside, this is coherent. The train receives the signalling and supervision information required by ETCS. ATO can then use the information provided by ETCS to compute and execute the driving strategy.

On lines not yet equipped with ERTMS/ETCS trackside, the situation becomes more difficult. The train may be equipped with ATO, but the standard signalling information required by the ERTMS/ATO architecture is not available through the usual ETCS chain. The line may still be operated with national signalling systems and lineside signals. These systems can support manual operation, but they do not directly provide the interoperable ETCS data expected by ATO.

This creates a paradox.

ERTMS/ATO is the target interoperable solution for automatic driving, but its use is constrained by the pace of ERTMS/ETCS trackside deployment. And ERTMS/ETCS trackside deployment will take time. It requires investment, engineering, authorisation, operational migration and coordination with rolling stock deployment.

The question is therefore not whether ERTMS/ETCS remains the target. It does.

The question is whether the railway can start preparing and using parts of the ERTMS/ATO architecture before every relevant line is equipped with ERTMS/ETCS trackside.

This is where migration concepts become useful.

A good migration concept should not create a new target architecture. It should not introduce a proprietary automation stack. It should not revitalise national Class B systems as the future. It should help the railway move towards the target while extracting some value during the transition.

Lineside Signalling Interpretation follows this logic.

Figure 1 — The migration problem

2. Lineside Signalling Interpretation as a migration concept

Lineside Signalling Interpretation is a concept for using existing lineside signals as an input to a train-centric migration layer.

The basic idea is simple.

A lineside signal displays an aspect. The train detects this aspect through an onboard perception function. The system determines which signal is relevant for the train, using localisation and infrastructure data. A signalling converter translates the interpreted signal aspect into ERTMS/ETCS-compatible information. This information is then provided to the ERTMS/ETCS onboard unit. Once the ETCS onboard unit has the necessary information, it can provide the standard input required by the ERTMS/ATO onboard function.

In simplified terms, the chain is:

the signal displays an aspect;
the train detects the signal;
the system interprets the aspect;
the aspect is converted into ETCS-compatible data;
ETCS supervises the train;
ERTMS/ATO can drive the train.

The important point is that the ERTMS/ATO onboard function remains the standard one. The migration concept does not change the ATO system into a national or proprietary automatic driving function. It changes the way the ETCS onboard unit obtains the information it needs in a non-ETCS trackside environment.

This is why Lineside Signalling Interpretation is interesting.

It does not compete with ERTMS/ATO. It tries to preserve ERTMS/ATO.

It does not replace ERTMS/ETCS. It tries to make use of the ERTMS/ETCS onboard unit earlier.

It does not define the final high-capacity railway architecture. It supports a transition period in which lineside signalling still exists.

This distinction is essential. If the concept were presented as a new target system, it would create confusion. The target remains radio-based ERTMS/ETCS where appropriate, ERTMS/ATO for interoperable automatic driving, and progressively more advanced DATO capabilities. Lineside Signalling Interpretation is not that target. It is a bridge.

The purpose of the bridge is to reduce the gap between today’s infrastructure and tomorrow’s architecture without creating another fragmented automation layer.

That is why this article treats Lineside Signalling Interpretation as a migration architecture.

Figure 2 — Lineside Signalling Interpretation: the basic chain

3. A train-centric emulation of ETCS Level 1

The concept becomes easier to understand when compared with ERTMS/ETCS Level 1.

In conventional ETCS Level 1, lineside signalling information is acquired on the trackside. A Lineside Electronic Unit is connected to the signal. It reads the signal aspect and selects the corresponding ETCS information. This information is then transmitted to the train through Eurobalises.

The functional logic is clear: interpret the signal aspect, associate it with ETCS data, and transmit that data to the onboard ETCS system.

Lineside Signalling Interpretation follows a similar logic, but moves part of the acquisition and interpretation chain onboard.

Instead of a trackside LEU and Eurobalises transmitting the information associated with each signal, the train uses onboard perception, localisation, Digital Map data and a signalling converter. The signal remains physically present on the infrastructure, but the chain that interprets its information becomes train-centric.

This is why the concept can be described as an onboard emulation of ETCS Level 1.

The expression is useful, but it should be used carefully. Lineside Signalling Interpretation does not mean that ETCS Level 1 is fully reproduced without trackside engineering. It does not remove the need for application engineering. It does not make signal interpretation trivial. The system must still know what each signal means, which track it applies to, which route is relevant, which ETCS information must be associated with each aspect, and how to manage inconsistencies or failures.

What changes is the allocation of functions.

In ETCS Level 1, the interpretation chain is mostly trackside.
In Lineside Signalling Interpretation, part of the interpretation chain is onboard.

This allocation shift is the core of the architecture.

It also explains why the concept raises important questions. If the train becomes responsible for detecting and interpreting lineside signals, the system must provide sufficient confidence in perception, localisation, map data, conversion logic and operational consistency. A camera reading a signal is not enough. The system must determine the correct target signal, understand the route context, associate the detected aspect with the correct infrastructure data and generate safe, consistent ETCS-compatible information.

This is not only computer vision.

It is system engineering.

Figure 3 — From trackside ETCS Level 1 to train-centric emulation

4. The system architecture

Lineside Signalling Interpretation relies on several building blocks. Each block is important, but the real challenge is the architecture connecting them.

The first building block is the ERTMS/ETCS onboard unit. This is a major point. The concept aims to keep the onboard train protection function aligned with the European target system. It does not replace ETCS with a national adapter or a proprietary ATP function. The train remains equipped with ERTMS/ETCS onboard.

The second building block is the standard ERTMS/ATO onboard function. The purpose is to avoid creating a specific ATO system for each national signalling environment. ATO should still receive its required information through the standard ETCS/ATO interface, as far as possible. From the ATO perspective, the migration layer should be invisible or at least contained.

The third building block is perception. The perception component detects the relevant lineside signal and reads its displayed aspect. This may involve computer vision, but the concept should not be reduced to image recognition. The real question is not only whether the system can see a light. It must know which signal is relevant, whether the detected aspect is reliable, and how the result should be used by the signalling conversion chain.

The fourth building block is localisation. The train must know where it is in relation to the infrastructure and to the target signal. Without localisation, perception cannot be interpreted correctly. A train may see several signals in the environment. Only one may be relevant for its route. The system must therefore combine perception with position, heading, route context and infrastructure data.

The fifth building block is the Digital Map. This map provides the infrastructure data needed by the onboard system: signal locations, track geometry, route information, signalling data and other elements required to associate a perceived signal with the correct ETCS-compatible information.

The Digital Map is not a convenience feature. In this concept, it becomes part of the signalling interpretation chain. If the map is wrong, outdated or inconsistent with the physical infrastructure, the interpretation may be wrong. Data quality, configuration management and update processes therefore become safety and migration issues.

The sixth building block is the signalling converter. Its role is to transform the interpreted signal aspect into ERTMS/ETCS-compatible information. It uses the perceived signal aspect, localisation, Digital Map data and, where available, operational information from traffic management or operational execution systems.

The seventh building block is the operational execution environment. The train still needs a mission. A Journey Profile may provide route and operational information. In some implementations, expected signal aspect information or operational context may also support the interpretation process.

Taken together, these components create a migration layer.

Existing lineside signalling remains on the infrastructure.
The train perceives and interprets it.
The Digital Map and localisation give context.
The signalling converter produces ETCS-compatible data.
The ETCS onboard unit supervises the train.
ERTMS/ATO uses ETCS information to drive.

The architecture is interesting because it keeps the target stack visible. The migration layer sits between legacy signalling and ETCS onboard, but the long-term objective remains convergence towards ERTMS/ETCS and ERTMS/ATO.

Figure 4 — Main building blocks

5. From signal aspect to ETCS-compatible data

The functional chain can be described step by step.

First, the train receives its mission. This may come through a Journey Profile or another operational data mechanism depending on the architecture. The train must know the route it is expected to follow and the relevant infrastructure context.

Second, the onboard system retrieves the required infrastructure data from the Digital Map. This includes the location and characteristics of the signals that may be relevant for the route.

Third, the train localises itself. It determines its position and heading with enough confidence to understand where it is in relation to the infrastructure and to the next relevant signal.

Fourth, the system identifies the target signal. This is a crucial step. Perceiving a signal is not sufficient. The system must know whether this signal applies to the train, to the route and to the track on which the train is moving.

Fifth, the perception component reads the signal aspect. It may detect whether the signal displays a restrictive aspect, a permissive aspect, a stop aspect, a warning aspect or another national signalling indication.

Sixth, the signalling converter uses the signal aspect, localisation and Digital Map data to select the corresponding ETCS-compatible information.

Seventh, this information is provided to the ERTMS/ETCS onboard unit. The ETCS onboard unit can then supervise the train according to the information it has received.

Finally, ERTMS/ATO can use the information provided by ETCS to perform automatic driving within the supervised envelope.

This chain shows why Lineside Signalling Interpretation is not just about reading a signal.

It is about converting a legacy signalling environment into a form usable by the European onboard architecture.

This also explains the importance of application engineering. The system must know how each signal aspect maps to ETCS-compatible data in its specific location and operational context. This mapping must be engineered, validated, configured and maintained. Temporary changes, infrastructure modifications, signal replacements or operational rule changes may affect the data.

The migration layer therefore becomes an engineered system, not a generic perception feature.

This is an important point for industrialisation. A prototype can demonstrate that a signal can be detected. A deployable railway system must demonstrate that the interpretation chain is safe, maintainable, certifiable and compatible with operations.

That is a much harder problem.

6. Migration benefits

The first benefit of Lineside Signalling Interpretation is that it preserves the standard ERTMS/ATO solution.

This is important because the temptation during migration is to create local shortcuts. If each country or operator develops its own ATO adaptation for its national signalling system, Europe risks reproducing the fragmentation that ERTMS is meant to solve.

Lineside Signalling Interpretation offers another logic. The migration layer may be specific to the transition, but the ATO onboard function remains aligned with the European standard architecture. This protects long-term interoperability and avoids turning migration into another source of fragmentation.

The second benefit is that it reduces the dependency on immediate ERTMS/ETCS trackside deployment.

ERTMS/ETCS trackside remains the target direction. But if ATO benefits can only be obtained after full trackside deployment, operators may have to wait many years before seeing value from automatic driving. Lineside Signalling Interpretation could provide an intermediate step: trains equipped with ERTMS/ETCS onboard and ERTMS/ATO could begin using automatic driving on lines still equipped with lineside signalling, provided that the operational envelope is clear and the safety case is acceptable.

The third benefit is that it may strengthen the business case for onboard ERTMS/ETCS.

One of the difficulties of ERTMS migration is synchronisation between trackside and onboard deployment. If trains are equipped before enough infrastructure is ready, operators may carry cost without operational benefits. If infrastructure is equipped before enough trains are ready, the network cannot deliver its full value.

A migration concept that creates earlier onboard value could help reduce this problem. If an equipped train can already use ERTMS/ATO benefits in selected non-ETCS trackside environments, the onboard investment becomes easier to justify.

The fourth benefit is preparation for higher automation.

Lineside Signalling Interpretation introduces capabilities that are also relevant for future automation: perception, localisation, Digital Map-based operation, onboard data integration, consistency checks, operational execution and train-centric processing of infrastructure information. These capabilities do not deliver GoA4 by themselves, but they belong to the wider family of functions needed for future DATO.

This is why the concept is interesting beyond GoA2.

It can support automatic driving earlier, but it can also help prepare the railway system for a more train-centric and data-driven architecture.

The key condition is to keep the migration logic clear. The concept should help the railway move towards the target, not create a new permanent detour.

7. Limits and open points

Lineside Signalling Interpretation must be treated carefully because it is not the final high-capacity architecture.

The first limitation is capacity. Because the concept is closer to an ETCS Level 1-like emulation than to radio-based ETCS Level 2 or Moving Block, it cannot deliver the same headway performance as the most advanced radio-based architectures. For high-density corridors, radio-based ETCS and future capacity concepts remain the target.

The second limitation is scope. The concept depends on the existence of lineside light signals that can be detected and interpreted. If a line does not use such signals, or if the signalling information is not visually available, another migration approach is required.

The third limitation is data quality. The Digital Map and application data become central. Signal locations, signal characteristics, track geometry, route information, ETCS conversion logic and operational constraints must be accurate and maintained over time. Data ownership, configuration control, certification and update processes become essential.

The fourth limitation is perception confidence. The system must detect the correct signal, read the correct aspect, assess the confidence of the result and manage discrepancies. Weather, lighting, occlusion, temporary signalling, unusual aspects, maintenance situations or unexpected infrastructure conditions may all create difficulties.

The fifth limitation is safety and certification. The concept changes the way signalling information reaches the ETCS onboard unit. That requires hazard analysis, safety arguments, failure mode analysis, operational procedures, degraded mode definition and clear responsibilities.

The sixth limitation is operational rules. The railway must define when the concept can be used, under which operating conditions, by which trains, at what speeds, in which areas, with which fallback modes and with which driver responsibilities.

These limitations do not invalidate the concept. They define the work needed to make it credible.

This is a normal pattern for railway migration concepts. The value of the idea lies in the architecture, but the feasibility depends on operational boundaries, engineering discipline and safety evidence.

Lineside Signalling Interpretation should therefore not be oversold.

It is not a magic shortcut to automation. It is a controlled migration layer that may be useful in selected contexts if its boundaries are well understood.

8. Architecture perspective

From an architecture perspective, Lineside Signalling Interpretation is interesting because it addresses a real migration problem without changing the target direction.

The target remains ERTMS/ETCS and ERTMS/ATO. The migration concept does not try to replace them. Instead, it inserts a layer between existing lineside signalling and the ERTMS/ETCS onboard unit.

This makes the concept a good example of technology infusion in a brownfield railway.

The brownfield system is the existing signalling infrastructure. It is still in operation. It cannot be replaced instantly. The target architecture is ERTMS/ATO operating within the ERTMS/ETCS safety envelope. The migration layer creates a temporary bridge between both worlds.

This is exactly the type of problem railway architecture must solve.

A good migration layer should provide value during transition, but it should not become the architecture itself. It should help the target system become useful earlier, not delay or fragment the target system.

This is the main architectural discipline required here.

Lineside Signalling Interpretation should not become a new national automation stack. It should not create a separate ATO product. It should not make the railway dependent on legacy signalling forever. It should not shift so much complexity onboard that certification, maintenance and operations become harder than the problem it solves.

It should be evaluated through three questions.

First, does it preserve the target ERTMS/ATO architecture?

Second, does it provide enough operational value during the migration period?

Third, does it reduce or increase long-term system complexity?

If the answer to the first two questions is yes, and the answer to the third is carefully managed, the concept may become a useful migration bridge.

If not, it risks becoming another layer in an already complex railway system.

This is why Lineside Signalling Interpretation must be assessed less as a perception feature and more as a migration architecture.

The challenge is not only to see the signal.
The challenge is to preserve convergence while making automation useful earlier.

9. Conclusion

ERTMS/ATO is the European interoperable solution for automatic train operation. It is designed to work with ERTMS/ETCS, which provides the protection and signalling information required by the ATO system.

This creates a migration challenge. ERTMS/ETCS trackside deployment will continue, but it will take time. Many lines will remain operated with national signalling and lineside signals for years. If ERTMS/ATO can only be used after full trackside deployment, automatic driving benefits may arrive too late for many operators.

Lineside Signalling Interpretation addresses this gap.

The concept consists in interpreting existing lineside signals from the train, converting their aspects into ERTMS/ETCS-compatible information, and feeding an ERTMS/ETCS onboard unit. ERTMS/ATO can then continue to operate through the standard architecture.

This can be understood as a train-centric emulation of ETCS Level 1. The trackside chain of signal, LEU and Eurobalise is replaced by a combination of perception, localisation, Digital Map and signalling conversion.

Its value is not to replace ERTMS/ETCS trackside deployment.

Its value is to support the transition.

It may allow earlier use of ERTMS/ATO, strengthen the onboard ERTMS/ETCS business case, introduce useful capabilities such as perception and Digital Map-based operation, and prepare some of the building blocks needed for future automation.

But the concept has limits. It does not provide the same capacity performance as radio-based ERTMS/ETCS. It depends on lineside signals. It requires robust data, localisation, perception confidence, safety analysis, operational rules and certification.

This is why it should not be treated as a target architecture.

Lineside Signalling Interpretation is a migration bridge between today’s lineside signalling and tomorrow’s interoperable automated railway system.

And in that sense, it is a useful example of how railway automation should be approached: not as an isolated technology, but as a system architecture problem.

Documentation and further reading

Voie Libre articles

• ERTMS: The European Rail Traffic Management System
• ERTMS/ETCS: The European Train Control System
• ERTMS/ATO: Europe’s Interoperable Train Autopilot

European R&D documentation

• TAURO project — Lineside Signalling Interpretation report
• TAURO project — Digital Map report
• Shift2Rail on ATO, perception, Digital Map, migration to ERTMS/ATO and future automation capabilities
• FP2-R2DATO work on Digital and Automatic Train Operation, GoA3/4, Remote Driving, perception