Even though railway technology advances, accidents still happen quite a lot; in the span of 2017 and 2021, in the United States, there have been 681 train accidents in total, causing countless casualties and a large amount of financial loss. The emergence and development of artificial intelligence (AI), the Internet of Things (IoT) and big data have provided the rail sector with an opportunity to forward into the future trains and railways in the aspects of AI behavior analysis, train diagnosis, and cybersecurity. As a leading supplier to the rail industry for decades, Axiomtek provides complete railway embedded solutions and leverages its field-proven technologies in AI, IoT, and edge computing to improve rail operations and maintenance.
AI Driver and Passenger Behavior Analysis
Although today the operation of the train is almost semi-automatic, the importance of the drivers does not degrade; the drivers still need to keep an eye on everything to ensure the safety of passengers and operations. As the drivers work alone, in case of being unfocused or any acute medical condition, with an edge AI system for onboard surveillance applications, the activity of the driver can be monitored and analyzed; for example, if the driver falls to the ground or stares at one direction for more than five seconds, the alarm will go off to alert the driver and the control center to evaluate whether to start the emergency procedures. In the same way, the abnormal or attack behavior of passengers can also be detected by the AI vision to ensure a safe ride.
Recently, our system has been even deployed in an original application, seat occupancy detection. Integrated with advanced software and cameras, the edge AI system can detect passengers sitting down or leaving their seats, providing real-time status data. For the convenience of the passengers, the occupancy of each car would be displayed in graphics on the electronic boards on the platform. More than that, the occupancy status can even be checked on mobile devices, which enhances and changes the scope of transportation.
Fueled by intelligence from AI-driven systems and applications, railway operations are becoming safer and smarter while improving passenger comfort. Axiomtek’s new UST510-52B-FL would be an applicable fit for light edge AI applications while it is powered by the Intel® Xeon® processor (Coffee Lake) with up to 65W TDP. Designed in a compact size, it has high-performance image processing computing power with two DDR4 system memory up to 64G, which can be further boosted with an accelerator module via M.2 or a mini PCIe slot.
What makes it a perfect fit for a behavior analysis system is that the system features eight PoE ports to support the IP cameras. Thus, the cameras can be deployed to multiple cars and run real-time surveillance. More than that, it is equipped with one M.2 Key B 3050 slot for the 5G module. Two hot-swappable 2.5-inch SATA drives with Intel® RAID 0/1 function and two mSATA are available for storage. The system has a rugged design that the vibration endurance is up to 3 Grms as SSDs are installed. Also, it has a wide operating temperature ranging from -40°C to +70°C to work under any harsh conditions. To keep the proof for unexpected incidents, Axiomtek suggests adding a Network Video Recorder (NVR) and thereby recommends the tBOX300-510-FL. The EN 50155 certified system has four swappable 2.5" SATA drives and supports RAID 0/1/10 to provide big storage space and data backup.
Onboard Diagnosis and Predictive Maintenance
To ensure a safe ride, we not only rely on the mechanics or the technicians to do the routine maintenance. With a variety of sensors such as sound sensors and the development of IoT, we advance to predictive maintenance which is able to detect possible upcoming malfunctions or maintenance in need. Through the signals and data analysis, train operators can know the status of the bogie or axle via software and tackle the problems as soon as possible. To obtain data from different sensors via various protocols, gateways with rich I/O are required. As gateways integrate data and transmit it to the central management system, the location and the type of problem can be quickly identified.
Axiomtek’s compact tBOX110-APL-TV is powered by the Intel Atom® x5-E3940 processor (Apollo Lake) and possesses rich I/O interfaces. Despite its small size, it features two CANBus and two COM ports supporting RS-232/422/485 with high-speed mode for parameter transmission without any latency; in addition to two LAN ports, it has two mini PCIe slots and external SIM card slots to enable a wireless connection. Two USB 3.0 ports and 6-in/2-out DIO make it ready for more peripherals, making it a cost-effective gateway to collect several digitalized signals.
The system supports DIN-rail or wall-mount to make the installation more flexible in the space constraint train cabin. Besides collecting data, when an emergency happens, the driver can send out warnings or evacuation announcements to the passenger information system (PIS). To withstand the challenging environments as railways reach the most distant places around the world, the tBOX110-APL-VT is able to perform under a wide range of temperatures from -40°C to 70°C.
The rolling stock sensors, devices, and systems are also in the scope of the operation technology (OT); as the OT converges with the information technology (IT) and provides data for insights in detail from the edge, the OT is hence exposed to the internet environment and under the threat of cyber-attack. The railway and related systems have been under attack around the world and caused chaos and disinformation. To prevent the trains from malfunctioning, installing a system to set up a firewall in the trains to protect systems such as the central control unit can be a helpful solution. The firewall can segment the OT network from the IT network and filter unknown packets.
The tBOX500-510-FL has been chosen by several system integrator customers and used as a hardware firewall due to its scalability. It is powered by the 7th gen Intel® Core™ processor and what makes it special is that it is equipped with two slots for a series of value-added modules to gain different I/O flexibility. To build up a railway cybersecurity solution, Axiomtek recommends installing two VAM708 modules to the system. Each module has four M12 A-coded GbE LAN ports, which means each system with two modules would have four bypass pairs. The modules not only provide multiple LAN ports to integrate signals from various onboard devices but also support LAN bypass to maintain the functions to keep the firewall and the whole management system working even when the system is down. Additionally, Axiomtek provides other modules supporting CANBus, PoE or serial, making the deployment flexible.
Keeping trains safe takes a lot of work in different aspects. In cyber security, AI and diagnosis systems, the applications all require different systems with diverse features. As all the systems coordinate and function accordingly, there will be a safe ride.