top of page

Software-Defined Train: Understanding ACTIA’s Railway HPC Demonstrator

Oct 30
3 min read

On September 8, 2026, ACTIA unveiled a railway demonstrator designed for the InnoTrans trade fair. The system integrates a central high-performance computer (HPC) with five ACTIVISION displays and four cameras. It illustrates an evolution in onboard computing: rather than relying on a multitude of specialized computers, various functions are consolidated onto a shared infrastructure—one where software can evolve throughout the long service life of a train.



Why Distributed Architecture Is Becoming a Hindrance


Modern trains incorporate functions for operation, supervision, diagnostics, communication, and monitoring. Historically, each of these has relied on dedicated electronic hardware and its own specific integration cycle. As the number of such functions grows, teams must manage an increasing array of hardware units, power supplies, networks, software versions, and interoperability constraints. While this model is robust in certain contexts, it lacks flexibility when an operator wishes to enhance the driver experience, deploy a new vision algorithm, or integrate a fresh interface. The core research and engineering challenge lies in determining which functions can be pooled—and at what level of safety—to achieve tangible gains across the entire product lifecycle.


Centralized HPC as a New Architectural Building Block


The demonstrator announced by ACTIA is built around an HPC (High-Performance Computer) unit, which consolidates processing tasks that were previously distributed. The manufacturer highlights a TCMS (Train Control and Management System) interface, video surveillance feeds, electronic rearview mirrors, and passenger detection or counting functions. The concept is reminiscent of certain "software-defined vehicle" platforms developed within the automotive industry. From an architectural perspective, the challenge lies in the simultaneous execution of multiple computational workloads, the controlled sharing of resources, and the decoupling of hardware from software functions. The inclusion of five screens and four cameras illustrates the convergence of data streams, though this alone does not prove that all certification requirements for a complete railway system have been met.


Key challenges to address: real-time operation, safety, and cybersecurity


Centralization does not mean that all functions are identical. Real-time requirements, criticality levels, and failure modes differ—for instance, between a passenger-comfort interface and a function linked to operational safety. It is essential to ensure isolation between applications, system availability, appropriate redundancy mechanisms, and predictable responses to faults. While remote updates offer flexibility, they also raise complex issues regarding validation, version traceability, and cybersecurity. Furthermore, the need to maintain rolling stock over several decades necessitates a long-term strategy for the sustainability of both hardware and software components. ACTIA has presented an architecture and a set of demonstration use cases; however, final performance and qualification will depend on the specific integration projects.


Innovation transferred from one sector to another


ACTIA explicitly cites the transfer of expertise from the automotive and industrial vehicle sectors to the railway industry. This serves as an interesting example of technology cross-pollination: a single computing unit can host vision-based and AI functions, but adapting it for a train requires a deep understanding of operational specifics, onboard networks, and industry-specific standards. In the context of technology monitoring, it is useful to observe not only railway manufacturers but also automotive computing architectures, hypervisors, isolation techniques, and the evolution of standards. Cross-referencing patents, articles, and demonstrators helps distinguish between proven technology transfers and concepts that remain purely speculative.


Perspectives for R&D Teams


A train with an evolvable architecture could more easily integrate new diagnostic tools, vision algorithms, or operational assistance functions. However, its commercial value will depend on maintenance efficiencies, the acceptability of interfaces, and the observed levels of system availability. GraphMyTech can support teams looking to map technologies related to "software-defined rail," identify R&D partners, and compare architectural trade-offs before launching a program.


 
 
 

Comments


Commenting on this post isn't available anymore. Contact the site owner for more info.

22 rue de Bagnolet,

75020 Paris, France

  • LinkedIn

Privacy & Cookies Policy

bottom of page