WELCOME to the New In Signal
by author Ivan Ristić...

Promo & Careers | Contact us directly for more info how to advertise for FREE!

Latest News

THE C-TAI™ CONTROL TABLE TOOL

The development of modern railway signalling systems increasingly requires automation, traceability, and formal verification of engineering outputs. One of the most labour-intensive tasks in signalling design is preparing control tables that define the conditions for route setting, switch positions, overlap requirements, signal aspects, and release logic for each route in a station. Traditionally, control tables are produced manually by signalling engineers, often requiring weeks of work and extensive verification. Their complexity grows exponentially with the number of tracks in a single station.
The Control Table AI tool (C-TAI™) has been developed to automate this process by transforming railway station layouts into validated, machine-readable control tables through a deterministic and auditable workflow.
The table below compares manual and automated preparation techniques.

From the estimated values above, it is very clear that such a tool development is quite necessary in order to outperform manual generation and to provide gains in case of very complex station layouts or in case several stations are processed in the same time period, both in terms of time spent and budget.

Figure 1. Architectural roadmap of the product

From CAD layout to control tables
The C-TAI™ tool begins with a railway station layout created in AutoCAD. The layout is converted from DWG/XML to DXF and parsed to extract signalling objects, such as tracks, signals, and switches. These elements are then transformed into a railway topology model, represented as a graph.
The extraction of topology is greatly enhanced in terms of accuracy and generalization by training hybrid vision-language models for element extraction and classification or ontology mapping to a unified schema. Graph-based reasoning models are then implemented to extract the underlying topology and flow from the parsed entities. This acts as a referee for the primary hard-coded extraction script to accurately deal with divergent layout structures and linguistic differences of terminologies.

Figure 2. The AutoCAD layout of the station

In this model:
• Nodes represent insulated sections
• Edges represent track connections
• Signals and switches are attached as attributes of the graph
From this undirected topology, the system generates two directed graphs that represent train movements in both directions (left-to-right and right-to-left) through the station. These graphs serve as the foundation for automated route analysis and enumeration.

Figure 3. The example of a graph representation of the CAD layout

Automated route enumeration and logic processing
Once the topology model is created, the C-TAI™ tool automatically identifies all possible routes between main/shunting signals. For each route, the system determines:
• the sequence of sections composing the route
• required switch positions
• potential overlap candidates
• intermediate signals encountered along the path
The tool then applies a declarative rules engine that incorporates specific (national) signalling rules. These rules define route-setting conditions, overlap policies, flank protection requirements, signal aspects, and release logic. Because the rules are defined in configurable rule packs, the tool can be adapted to different national signalling systems without modifying the core algorithms.
Using fine-tuned Language Models on small, annotated country-specific datasets, cross-country signalling logics can be encoded in a pre-determined and structured format, allowing rule-based validation and interlocking logic to adopt multilingual and multi-demographic robustness under one ecosystem.

AI-assisted route editing and topology generalization

In addition to deterministic route enumeration, the C-TAI™ tool can also incorporate an AI-assisted support for route editing, topology generalization, and engineering refinement of automatically generated data. After the CAD layout is parsed and transformed into a graph-based topology model, the system applies pattern-recognition algorithms to identify typical signalling structures such as crossovers, sidings, and platform tracks. These structures can be further generalized into reusable logical patterns that allow the tool to interpret engineering intent rather than only raw geometry.
Through this mechanism, the system can automatically group related routes, detect equivalent route patterns, and simplify large station layouts into structured route families. Engineers can then modify, merge, or split routes through AI-supported editing functions without altering the underlying CAD layout. This capability significantly accelerates the preparation of control tables in complex stations where hundreds of routes share common infrastructure elements. The AI, therefore, acts as an engineering assistant, enabling rapid adjustment of route definitions while maintaining full traceability between the original CAD topology, the generated logical model, and the resulting control tables.

Generation of control tables

Based on the enumerated routes and rule evaluation, the C-TAI™ tool automatically synthesizes control table entries. Each row of the generated control table includes:
• route identifier and direction
• start and destination signals
• ordered list of track sections
• required switch positions
• overlap and flank protection conditions
• signal aspect logic
• route locking and release logic
The results can be exported in several formats, including CSV, XLSX, PDF, JSON, and optionally XML (railML), enabling integration with engineering workflows and other signalling design tools.

Figure 4. The example of the generated control table

The C-TAI™ also includes a dedicated front-end environment designed to provide signalling engineers with an intuitive interface for reviewing and managing the automatically generated results. Within this environment, all signalling objects extracted from the CAD layout, such as track sections, switches, signals, and routes, are represented as vector graphics that remain linked to the underlying logical model. This enables engineers to interact directly with the digital station layout, selecting individual objects or routes and instantly viewing their associated control table conditions.
Any edits or rewrites, whether AI-based or manually from the front-end, will be stored as pointers to different branches, with the user’s metadata, changes made from the original, date, etc., in a tree/graph-like structure, as in version control systems like Git. Users will also have different roles (like admin, editor, viewer) based on their accessibility and control over editing and deletion.
The interface also supports control table selection and inspection through an HMI-like visualization of route states and infrastructure elements. For example, when a specific route is selected, the system can highlight the corresponding track sections, required switch positions, overlap zones, and conflicting routes directly on the graphical layout. This approach mirrors the operational logic used in real interlocking HMI and allows engineers to verify route conditions in a visually intuitive way. The integrated ecosystem, therefore, combines CAD-derived geometry, AI-generated logic, and engineering validation tools within a single ecosystem, enabling efficient modification of control tables while preserving consistency between the graphical layout and the logical signalling model.

Figure 5. Example of the front-end detail layout (HMI-style visualization)

Figure 6. Example of the front-end detail layout (control table details)

Verification and engineering benefits

The C-TAI™ workflow also includes automated verification and validation steps, such as conflict detection and route exclusivity checks. These mechanisms help detect errors early and ensure the generated control tables are consistent with signalling rules.
At the same time, a human-in-the-loop approach remains an essential part of the system, building confidence and trust among railway professionals by preserving reliability and safety standards, while reaping the benefits of automation. Railway signalling tables have traditionally been created and validated manually by experienced signalling engineers or officers, and operational safety demands that domain experts retain full oversight of any automated process. For this reason, the tool is designed to allow railway officers and signalling engineers to review, modify, and rewrite control table entries directly from the frontend interface whenever required. This ensures that any ambiguity, local operational rule, or site-specific constraint can be incorporated easily.
Compared with traditional, fully manual processes, the C-TAI tool offers several advantages:
• significant reduction in engineering time
• deterministic and reproducible outputs
• full traceability between layout, rules, and generated tables
• easier adaptation to different national signalling systems
• improved support for digital safety cases and further automated verification

Table 1. A brief comparison with traditional manual preparation process

Toward digital signalling

The C-TAI™ tool represents an important step toward digital transformation in railway signalling engineering. By converting infrastructure layouts directly into machine-readable logical models and automatically generating control tables, the tool supports the transition toward model-based design, automated verification, and digital safety cases.

As railway systems become increasingly complex and safety requirements continue to grow, such automation tools will play a crucial role in improving efficiency, reliability, and transparency in signalling design processes.

If you are interested in finding out more details, please contact us at the following addresses:

valerio.divico@sinauragroup.com

artur.wolnica@awr-engineering.com

ivan.ristic@signalling-solutions.com