Just in time for the railway fair InnoTrans in Berlin, Syslogic is presenting its AI Rail Computer. Self-driving trains are already in use today, but often on limited networks. In the future, self-driving trains could also be implemented on complex railway networks. The simple fact that trains run on rails reduces the environmental influences and choices that must be made in it. Besides the obvious key goal of self-driving trains, there are several sub-applications in which AI and its associated disciplines of deep learning and machine learning can make an important contribution.
Syslogic is one of the leading companies in embedded systems for use in trains. The robust, durable and long-lasting industrial computers are used as rail control computers, data loggers, IoT gateways, and in PIS computers (passenger information system). Syslogic works with well-known railway companies. AI offers enormous potential for both vehicle manufacturers and railway operators. Accordingly, Syslogic and its partners have invested their efforts in developing an AI railway computer.
Syslogic uses a Jetson TX2 module by Nvidia as the basis for the AI Rail Computer. This module features an economical but powerful quad-core processor platform. The core of the Jetson TX2 is the ARM SoC Tegra X2 named Parker. It combines two computation modules with Nvidia's proprietary Denver 2 microarchitecture with four Cortex-A57 cores and a Pascal GPU. The latter has 256 shader cores. This enables the rail computer to use artificial intelligence in real time. The innovative processor technology combines Syslogic with the proven features of existing railway computers. For example, the AI Rail Computer is passively cooled, has a robust housing and screw-on M12 connectors. Thus, the AI Rail Computer complies with railway standard EN50155, class OT4, as well as fire protection standard EN45545. Furthermore, the railway computer has a galvanically isolated nominal power supply of 24 to 110VDC.
Besides performance and robustness, Syslogic is also focusing on easy commissioning of the AI Embedded Computer. Therefore, the devices are delivered with a preinstalled Ubuntu Board Support Package. Additionally, a complete developer environment including CUDA libraries is installed. The Nvidia software development kit JetPack including a deep learning framework for easy implementation of applications are available for a quick introduction to the AI world. For system connection, the AI computer features a flexible interface layout and WiFi, GPS and LTE functions can be added. Syslogic is currently providing their AI hardware platform as prototypes. The first few applications will soon be tackled with existing railway transportation customers.