Fueling Future Needs
Providing solutions for everything from autonomous operation to traffic management to electric vehicle infrastructure, embedded PC vendors are rolling out a variety of box-level systems integrating AI, vision technologies, 5G comms and more.
There’s no doubt that the “intelligent” part of today’s intelligent transportation system has moved to a whole new plateau. System designs are relying on advanced embedded PC technology to provide sophisticated networking, data collection and real-time control, all while meeting safety standards. The idea of employing artificial intelligence (AI) is no longer a niche idea for intelligent transportation, but more a routine requirement.
Vehicles of every kind—buses, public safety vehicles and truck fleets—are all embracing advanced technologies. For trucking fleets, centralized fleet management systems for fleet managers, real-time vehicle communication and live vehicle route tracking all need advanced computer and wireless communications. And smart buses are integrating public information systems, driver monitoring and in-vehicle safety monitoring. Even heavy construction vehicles are using embedded computing for safer more efficient operations.
To enable all these current and next-gen transportation needs, embedded computing systems vendors have kept pace, rolling out embedded PC solutions to meet a variety of design needs. Over the last 12 months, vendors have released a variety of box-level systems designed for both for in-vehicle use and transportation infrastructure systems.
MEETS RUGGED STANDARDS
A large subset of embedded PCs is targeted for a broad set of markets, some including compliance to multiple application-specific certifications. Along just such lines, in September OnLogic rolled out its Karbon 400 series of rugged computers. The two fanless, rugged customizable computers are powered by Intel Atom x6000E series processors, formerly known as “Elkhart Lake.”
In addition to meeting EN 62368-1 safety requirements, the Karbon 400 Series passed extensive EMC testing to EN 60601-1 medical, EN 50121 railway, EN 55032/35, Automotive E-mark and EN 60945 maritime standards (Figure 1). With 15kV ESD protections, 30A/m magnetic immunity and RF immunity up to 30V/m, the Karbon 400 Series has been engineered from the ground up by OnLogic for reliability. The systems also passed testing to IEC and MIL-STD-810 shock and vibration standards.
The Karbon 400 Series includes two models. Both the Karbon 410 and Karbon 430 can be configured with either the Dual-Core Intel Atom X6211E or Quad-Core Intel Atom X6425E CPU, which exhibited a 2x improvement in performance during OnLogic benchmarking. Built-in modularity allows users to choose the connectivity and capabilities needed for their application.
With an operating temperature range of -40°C to 70°C, fanless cooling, variable power input and zero moving parts, the Karbon 400 Series has been engineered to thrive in even the most challenging computing environments. The Karbon 430 model adds two additional M.2 2280 expansion slots, which can be used to equip additional storage, wireless connectivity or I/O. The Karbon 430 also offers an additional SIM card slot, enabling multi-carrier cellular connectivity configuration.
AI-ENABLED IN-VEHICLE COMPUTING
AI-based vision and machine learning have become important technologies for vehicle computers. Along such lines, in November DFI announced its VC300-CS, a fanless and AI-enabled in-vehicle computer. Powered by Intel 9/8th Gen Core processors and integrated with an Nvidia MXM module, VC300-CS accelerates graphic processing for AI vision and machine learning (Figure 2). In addition, 5G capability brings up real-time communication and response to the dynamics at the edge. VC300-CS achieves tremendous performance for advanced driver assistance, risk predictions, and onboard surveillance, while also delivering passenger comfort and safety in today’s increasingly diverse ecosystem of intelligent transportation, says DFI.
On the moving vehicle, AI vision and machine learning require abundant images and data for precise detection and analysis, making GPUs even more essential for future vehicles. Providing processing capabilities of up to 150W, 3072 CUDA cores, and 64TFlops by the MXM3.1 module, VC300-CS is powerful enough for autonomous driving workloads. Its compact and rugged design plus better heat dissipation, makes it a good choice for in-vehicle computing.
To meet future transportation needs, it’s important to be able to transfer and analyze a large amount of data in real-time. Therefore, the 5G network, featuring multi-gigabit speeds and ultra-low latency, strengthens the synergy of devices and makes communications seamless and efficient to achieve immediate roadway hazard detection and rapid response in emergencies.
VC300-CS not only supports four 15W M12 X-coded PoE, 2kV isolated DIO and numerous ruggedized I/O interfaces but also provides multiple M.2 and mini PCIe slots for LTE, 5G, Wi-Fi, GNSS, CAN bus or MVB functions. In addition, four external 2.5” SSD bring abundant data storage and disk arrays function. The comprehensive I/Os meet varying system requirements in smart transportation including wayside or onboard applications.
VC300-CS supports a wide range 9VDC to 36VDC power input to protect it from overcurrent and overvoltage. To guarantee the in-vehicle experiences, VC300-CS supports Power Ignition to shut down the system correctly and avoid reducing the computer’s performance and lifetime when the engine power cuts off. Since vehicles always face a huge temperature difference outdoors, VC300-CS has excellent heat dissipation and can operate in a harsh environment of -25°C to 70°C even if deployed under the driver or passenger seat without active fans.
TRAFFIC MANAGEMENT APPLICATION
Cincoze offers an application story about a system that uses its DS-1200 embedded computer. Cincoze’s customer provides ICT (information and communications technology) services and develops smart city solutions in Western Asia through the development of intelligent transportation for metropolitan municipalities. They deploy mobile detection systems to capture and process images of vehicles for license plate recognition. The detection equipment in the system automatically detects violations in the area. A robust in-vehicle computer is required to accurately record, process and transmit data.
Since fast image processing, live recording and automatic number plate recognition are the main applications of this system, the customer requires high-performance computing power without latency, providing immediate inference analysis and event response for law enforcement, while capturing high-resolution video and images from connected cameras. All the equipment in the mobile detection system has passed shock/vibration tests. The computing system needs to be tough and reliable to operate in harsh and extreme environmental conditions. The computer has to be I/O rich to connect to many devices, including cameras, IR lighting, sensors and peripheral security devices, to detect vehicles violating the rules. Easily replaceable additional I/Os are necessary for maintenance. Moreover, leveraging 4G LTE wireless connectivity for continuous connectivity with the cloud is required.
With all those requirements in mind, the system designers chose Cincoze’s DS-1200 series embedded PC. It supports an Intel 9th/8th Gen Core, Pentium or Celeron 35W/65W processor and supports DDR4-2666 SO-DIMM memory (Figure 3). With additional video capture cards or GPU cards, the DS-1200 allows for fast, accurate real-time visual inference analysis.
The DS-1200 series operates in a broad temperature range of -40°C to 70°C, ensuring reliable operation in harsh environments. It uses industrial-grade components and features a fanless and cableless design, a wide-range DC power input (9VDC–48VDC), a high tolerance for vibration/shock (5G/50G), and industrial-grade protection (OVP, OCP, ESD surge and so on). It has passed multiple rigorous tests and complies with industry standards EN50155 (EN50121-3-2) for reliable operations in moving vehicles or railway environments.
The DS-1200 series comes with a rich variety of I/O connectors, including 2x GbE LAN, 8x USB, 1x DVI-I, 2x DisplayPort, 2x RS-232/422/485 and a remote power/reset connector. The DS-1200 offers 2x SATA III bays, 3x full-size Mini-PCIe slots and 1x M.2 M-key 2280 socket for NVMe. The system also has 2x front-accessible SIM card slots for 4G LTE connections. Through Cincoze’s CMI/CFM expansion, the DS-1200 supports up to 8x GbE/PoE+ ports and more I/O or other functionality such as power ignition sensing. Two other models, the DS-1201 and DS-1202, have one and two PCIe expansion slots for add-on card expansion.
ENABLING AUTONOMOUS VEHICLES
According to Lanner, the combination of 5G wireless data networking and edge computing provides the ability to timely handle, process and analyze large amounts of data, which leverages the full potential of advanced technologies such as autonomous vehicles and the Internet of Things (IoT). Autonomous vehicles will change the experience of riding in cars, making it more pleasant, less stressful and more productive.
Autonomous vehicles rely on sensors, machine learning systems, complex algorithms and powerful processors to gather and analyze the data in order to send instructions to the car’s actuators, which control acceleration, braking and steering, says Lanner (Figure 4). To meet the unique needs of an autonomous vehicle system, the platform had to meet specific requirements. This includes high-performance computing power and supports server-grade Intel Xeon Scalable processor with extreme performance. 10Gbit Ethernet is also needed for high uplink and downlink bandwidth for inter-networks communications.
To meet those requirements, Lanner offers its NCA-5520 Intel Xeon edge computing hardware solution for autonomous technologies. Powered by a 2nd gen Intel Xeon processor scalable family and Intel C626 or C621 chipset, the Lanner edge computing platform features optimized computing power and virtualization capacity in a compact 1U form factor.
The edge computing platform supports improved memory bandwidth with up to 320GB DDR4 system memory at 2933/2666/2400/2133 MHz (by CPU), and flexible HDD/SDD storage options. The platform supports PCIe expansion of GPU accelerator for machine learning and processing, in addition to 10G fiber connectivity. The edge computing platform also provides advanced networking features for maximizing packet processing efficiency and cryptography acceleration.
COLLABORATION FOR C-V2X
For some embedded PC vendors, teaming up with other companies provide a means to an autonomous vehicle technology solution. Along those lines, in October, ADLINK Technology announced teaming up with ecosystem partners Askey Computer, the Institute for Information Industry (III) and Excellence Opto. Inc. (EOI) to develop and deploy an end-to-end C-V2X (C-V2X (cellular vehicle-to-everything) solution.
The collaboration integrates AI-enabled technologies and applications with 4G LTE/sub-6GHz 5G NR vehicle network connectivity, smart traffic lights and signal controllers. The C-V2X solution is expected to significantly improve road safety and traffic efficiency, establishing a solid foundation to help accelerate technology innovation and commercialization of intelligent transportation systems (ITS), autonomous driving and connected vehicles.
V2X technologies are designed to connect vehicles to each other (V2V), to pedestrians (V2P), to roadway infrastructure (V2I) and to the network (V2N). The 802.11p Wi-Fi standard established the foundation for V2X communications to enable drivers to avoid accidents and traffic jams, says ADLINK. With a wider range than a dedicated short-range communication (DSRC) and much lower latency, 4G LTE and 5G based C-V2X will communicate real-time data further. And, when used together with onboard intelligence, it will accelerate more intelligent transportation systems (ITS) and Level 3 and above autonomous driving.
The collaboration addresses the complex technology challenges from data collection to communication to processing in real-time by leveraging ADLINK’s expertise in developing and deploying rugged AI-enabled in-vehicle computers, private 5G networks, 5G edge servers and intelligent edge infrastructure. The C-V2X solution integrates ADLINK’s AI-enabled in-vehicle computer platform based on the AVA-3510 series, with Askey’s onboard/roadside unit (RSU/OBU), III’s AI-enabled applications including image-based AI vehicle identification, tracking and warning technology and EOI’s smart traffic light and traffic signal controller (Figure 5).
There are already four C-V2X pilot projects underway across Taiwan that implement both direct sidelink and cellular network communications. The projects are designed to test various use cases based on the enhanced 5G quality of service (QoS) for ultra-reliable, low-latency and high bandwidth communications.
GPUs FOR INTELLIGENT TRANSPORTATION
For its part, Aetina has placed its bets on Nvidia GPU technology to address the needs of transportation and transportation infrastructure system designs. According to the company, the ability of GPUs to optimize parallel computing results high performance and low power consumption for more significant and more extensive deep learning in neural networking. Figure 6 shows how an embedded Nvidia-GPU platform provides a solution across several transportation systems.
Among Aetina’s latest Nvidia GPU-based systems is its DeviceEdge Mini Series. The product line starts with three embedded mini-PCs, each of which run Ubuntu 18.04 on a choice of Nvidia’s Jetson Nano, Jetson Xavier NX or the Jetson TX2 NX. Two Intel-based Aetina AI inference systems were revealed early last year that support Nvidia graphics cards: a SuperEdge system with a Xeon-D and support for Nvidia T4 cards and a MegaEdge with Coffee Lake and GEForce RTX-3900 support.
The DeviceEdge Mini Series supports edge AI applications in smart transportation, factories, retail, healthcare, robotics and other AIoT use cases. The DeviceEdge Mini M1, M2 and M3 models have similar footprints and features. The M1 version adds 1x PoE PD 802.3at port, enabling system to be powered via Ethernet. Meanwhile, the M2 version adds 2x PoE PSE 802.3af ports, enabling system to power up to 2x PoE cameras at up to 100M. And, finally, the M3 version adds 6x additional USB 3.1 Gen1 ports (8x total) for powering up to 8x USB cameras.
5G FOR VEHICLE SYSTEMS
For its part, Sintrones also sees 5G as a critical part of the technology picture for in-vehicle computing. The requirements of smart applications center around AI, says Sintrones, and with the new lineup of 5G communication modules and GPUs, in-vehicle computers are on the same track with the future development of next-gen electric vehicle applications.
According to Sintrones, in order to keep up with the advent of 5G applications, the company’s EBOX-7000 series has built-in hot swappable SIM card auto activation modules, which offers Dual SIM Dual Standby, allowing end users to easily switch between SIM cards. The EBOX-7000 series also supports different 5G telecom service providers, offering users the best online experience.
The EBOX-7000 edge computer is powered by Intel 9th Gen Core i7/ Intel 8th Gen Core i7/i5/i3 CPU with 6x RJ45 GbE (optional 4x PoE Max. 100W). EBOX-7000 features TPM 2.0 and memory 2x DDR4 2400/2666MHz SO-DIMM up to 64GB (Figure 7). Furthermore, the environmental tolerance continually maintains a wide range of operating temperatures (-40ºC to 70ºC). In addition, it provides Intel UHD Graphics 630 with max resolution (HDMI) 4096×2304 at 24Hz.
Looking at its range of vehicle system customers, Sintrones’ sees several that serve as indicators. Aside from German snowplow vehicles, the application in AI smart buses in Singapore is a key milestone for Sintrones’ high-grade model. This case combines AI with unmanned autonomous vehicle systems by integrating ADAS and using GPUs’ high-speed computing capabilities to predict road conditions. At the same time, it meets the various needs of vehicle emergency response; it is currently one of the best models in high-end applications.
Sintrones AI applications have also entered the Japanese market. Currently, a key example is a case with a client who specializes in the manufacturing of excavators and construction machinery, where Sintrones’ systems are used as integrated applications in machinery safety warnings and GPS linkage, says the company.
VEHICLE NVR AND SURVEILLANCE
Video recording and surveillance are key parts of today’s transportation fleet management systems. Serving such needs, in November Axiomtek introduced the UST510-52B-FL, its E-Mark certified in-vehicle embedded system with a maximum of eight RJ-45 PoE connectors. This all-in-one box PC is powered by the LGA1151 Intel Xeon, 9th/8th generation Intel Core i7/i5/i3, Celeron or Pentium processor (codename: Coffee Lake) with up to 65W TDP and has the Intel C246 chipset (Figure 8). Its Smart Ignition power control prevents and protects the vehicle’s battery from unstable voltages. The UST510-52B-FL is well suited for in-vehicle applications including NVR (network video recorder), onboard security surveillance, vehicle controls, fleet monitoring and onboard passenger infotainment systems.
The UST510-52B-FL can support eight RJ-45 GbE PoE outputs with its internal PoE PSU solution, which generates 90W of PoE power. IP camera compatibility allows users to connect directly without installing an extra PoE switch, minimizing overall deployment costs and installation space onboard. Two additional HDMI and VGA ports enable up to three surveillance monitors to provide drivers and commuters with timely and accurate on-road and passenger safety information. Moreover, the system supports a wide temperature range of -40°C to 70°C and vibration of up to 3Grms. Two variable power modes result in an efficient, durable system. The UST510-52B-FL supports 12VDC and 24VDC typical automotive power input with Smart Ignition or a wide-range 9V to 36V DC power input.
Axiomtek’s UST510-52B-FL comes with two 260-pin DDR4-2666 SO-DIMM slots for up to 64GB of non-ECC/ECC memory. For expandability and versatility, the fanless embedded system offers one full-size PCIe Mini Card slot with support for mSATA/USB/PCIe and one half-size PCIe Mini Card slot with support for mSATA/USB. It also features one M.2 Key E 2230 for Wi-Fi and Bluetooth modules, one M.2 Key B 3050/3042 for 4G LTE, and two SIM card slots for 5G radio connectivity. For additional storage, two hot-swappable 2.5″ SATA drives with Intel RAID 0/1 function and two mSATA are available.
The UST510-52B-FL supports rich I/O interfaces including eight GbE PoE, four USB 3.0, two RS-232/422/485, one isolated 6-in/2-out DIO, one VGA, two HDMI, one audio (Mic-in/Line-out) and four antenna openings for WLAN and WWAN usage. Two flexible I/O windows for CAN Bus, LAN, DIO as well as Audio modules are included. The UST510-52B-FL also offers Trusted Platform Module 2.0 (TPM 2.0) to ensure critical information security and is compatible with Windows 10 and Linux.
RUGGED IN-VEHICLE SYSTEM
Because industrial systems and in-vehicle applications have several overlapping requirements, it’s not uncommon for embedded PC product lines to target both with some product lines. As an example, in September Aaeon introduced the VPC-5620S industrial embedded system. Available in two builds, the slim VPC-5620S IS and in-vehicle VPC-5620S VS, the system leverages a modular design to combine the power 8th Generation Intel Core processors with greater performance, storage capabilities and expansion support.
The VPC-5620S brings the performance of the 8th Generation Intel Core processors (formerly Whiskey Lake) to a rugged industrial platform designed to power embedded AI Edge applications anywhere they’re needed. With up to 64GB of memory, the VPC-5620S can power intelligent visual analysis applications such as Smart Factory and Security, and with expansion slots, can support AI accelerators to scale up the inference processing capabilities.
Available in two standard configurations, the VPC-5620S IS industrial system and VPC-5620S VS in-vehicle system (Figure 9), this versatile platform offers broad I/O features including four Smart PoE PSE ports designed for PoE cameras. Built to simplify deployment of visual applications, the Smart PoE ports can be managed with a user-friendly interface, enabling users to control voltage outputs or turn ports on and off remotely. Additional I/O features include four USB3.2 Gen 1 ports, two COM ports, HDMI and 8-bit DIO. Thanks to the modular design of the VPC-5620S platform offers additional configurations available on a per-project basis.
The VPC-5620S offers expandability to help quickly add on functionality and storage flexibility. Wireless connectivity is supported through two mPCIe slots, designed for Wi-Fi and LTE network deployments, along with an optional M.2 slot for 5G support. The system offers and M.2 slot with NVMe support for fast read/write speeds, as well as two 2.5″ SATA drive bays in standard configuration to support local network storage needs, such as for NVR and AI Surveillance applications. Users can also utilize eMMC and mSATA storage options.
One key feature of the VPC-5620S is its rugged design. Built to operate in a wide range of environments, the system boasts an operating temperature range of -20°C to 70°C. The system is tested to MIL-STD-810G shock and vibration standards, and the fanless design helps keep dust and contaminants out, ensuring long-lasting, consistent and reliable operation. Additionally, the VPC-5620S comes standard with wide voltage input, supporting 12V to 24V inputs.
SOLUTION FOR EV CHARGING STATIONS
Not all embedded transportation design is about computing inside the vehicle. Infrastructure systems like electric vehicle (EV) charging stations are an important part of the puzzle as well. With just that in mind, Vecow describes a design win for its IoT Gateway VIG-120 in an application as a comprehensive solution for EV charger management (Figure 10). Powered by an i.MX6 Arm Cortex-A7 processor, Vecow’s fanless VIG-120 integrates with seamless connectivity including 4x COM, 2x CANbus, 2x GigE LAN, 1x Mini PCIe and SIM socket for expansions, a wide DC input range of 6V to 40V and -25°C to 70°C operating temperature.
The customer was looking for an embedded system that is capable of providing efficiency, reliability and convenience for EV charger management and equipment monitoring. A reliable and smarter EV charger plays a large role because it leads to pleasant user experiences for all EV drivers, whether they are home, or needed for a long trip. In addition, the system needed high performance for real-time monitoring and control. It is designed for industrial environments and allows a non-stop communication network.
The Vecow VIG-120 is a programmable IoT Gateway to implement with third-party open-source database software like MySQL/SOLite for real-time data management in the field side. It has both 2x GigE LAN ports and 4G/Wi-Fi/NB-IoT/LoRa wireless communication expansions for the data transmission and back-up to the host of information management center. The platform also provides 2x CAN Bus ports with half-duplex, daisy chain, quick transmission rate, anti-interference and error management features to access with EV charger stands for CAN protocol conversion. For EV charging field environment monitoring, the VIG-120 is outfitted with 4x software configurable COM RS-232/422/485 ports, which enables the connection of the device to different front-end sensors or meters.
HEAVY EQUIPMENT SAFETY SYSTEM
In September, VIA Technologies showcased its VIA Mobile360 Heavy Equipment Safety System at MINEXPO 2021 in Las Vegas. The system leverages VIA’s 360-degree video and mmWave radar smart sensor fusion technology to provide operators of mining, excavation and construction vehicles with real-time alerts of approaching vehicles, machinery and workers in all weather, lighting and working conditions. Including a ruggedized IP67 waterproof enclosure, IP69K cameras and mmWave radar sensors, the system can be installed on a wide variety of loaders, off-highway trucks, and other heavy equipment.
The VIA Mobile360 Heavy Equipment Safety System is a complete solution that includes the VIA Mobile360 M810 System with an IP67 ruggedized waterproof enclosure, four FOV 190-degreee IP69K 720p cameras, three IP69K 77GHz mmWave radar sensors, a choice of 8.2″ and 10.1″ HD monitors, and a 4.4W speaker. An extensive range of mountings and cables is also available. To provide operators with a clear picture of everything that is happening around their vehicle, while eliminating blind spots and potential collision points, the VIA Mobile360 AI Mining Kit (Figure 11) supports a host of smart visual intelligence technologies, including VIA Mobile360 SVS and VIA Mobile360 ADAS.
To ensure the accurate and timely detection of potential safety hazards in all weather, lighting and operational conditions, the VIA Mobile360 Heavy Equipment Safety System features smart sensor fusion technology combining camera detection with mmWave radar sensors. A single, unified interface on the in-cabin screen featuring real-time 360-degree video makes it easy for operators to view alerts and see everything that is happening around their vehicles.
The system features two configurable Warning and Critical detection zones to notify operators of potential hazards around the left, right and rear sides of the vehicle. Whenever it detects an object or a person in either zone, it displays a color-coded indicator on the operator’s in-cabin screen. If it deems the object or person as a threat due to their trajectory, a blue box will flash around the color-coded indicator and the corresponding audio alert will also be played to alert the operator of the impending danger. When the vehicle is reversing, all objects detected by the system’s rear camera and radar are treated as threats regardless of their trajectories, enabling the operator to take extra care when carrying out this most collision-prone maneuver.
EMBEDDED DISPLAY INCLUDED
When used as a user interface for the driver, it’s handy for an in-vehicle computer to include an embedded display. As an example, in November Nexcom announced a rugged vehicle mount computer with an 8″ touchscreen for vehicle fleets, port warehouse management, vehicle control in mines and indoor and outdoor stacker storage vehicles used for smart warehousing and logistics management. The VMC-220 runs Linux on a quad-core, Cortex-A53 i.MX8M clocked at 1.3GHz. Android is available by request (Figure 12).
The 2.4kg, 250mm × 179mm × 67.5mm system is protected per IP65. The bumper guard protected computer also meets IK08 impact standards and is protected against shock and crash hazard per MIL-STD-810G, 516.6 and vibration per MIL-STD-810G, 514.6 (2g at 5Hz to 500Hz). Nexcom engineers were even emboldened to drop a 1.7kg steel ball on top of it, and the system appears to have withstood the abuse.
The fanless, VESA mountable VMC-220 supports -40°C to 70°C temperatures, given proper airflow and offers 10% to 90% (non-condensing) relative humidity tolerance. There is a very wide-range, M12 connected 9VDC to 60VDC input with voltage protections, ignition controls and a power button. An optional back-up battery with 20-minute duration is helpful for forklifts that need to change batteries, says Nexcom.
The 8″, IPS touchscreen has 1280×720 resolution, 1000:1 contrast ratio and 170-degree viewing angles. The backlit, anti-glare display has an exceptionally bright 1000 cd/m² level for sunlight readability. The projected capacitive touchscreen is 3mm thick and supports gloved input. It even works when there is water on the screen, claims Nexcom. The VMC-220 is equipped with 4GB of 2400MHz LPDDR4 plus 32GB eMMC, with a BOM option to go to 64GB. You also get a covered microSD slot.
Camera interfaces and other vision technologies are top requirements in many in-vehicle system designs. For its part, Eurotech offers its BoltGPU 10-31, a rugged, fanless Subsystem for AI and machine vision at the edge. With rolling stock and automotive certifications, and multiple connectivity options, it enables power efficient, GPU-accelerated workloads in harsh operational environments. The BoltGPU 10-31 is powered by NVIDIA Jetson Xavier NX and combines a 6-core, high-performance Arm CPU with a 384-core GPU and 48 Tensor Cores, offering exceptional power efficiency and up to 21TOPS of accelerated computing (Figure 13).
Fanless and very rugged, the BoltGPU 10-31 allows the simultaneous execution of neural models and the processing of multiple high resolution, high frame rate sensors, making it the ideal platform for AI and machine vision workloads in harsh environments. The BoltGPU 10-31 has 16GB of eMMC that can be further expanded with an internal, high speed NVME that provides additional storage for video streams and sensor data. A 4-position mini FAKRA connector provides up to 4x GMSL camera inputs. Three GbE interfaces offer fast connectivity to the network and to IP cameras. USB3 Cameras can also be connected, thanks to dual USB3.1 Gen 2 ports (noise and surge protected). Other interfaces include isolated CAN-FD, opto-isolated DIO and video out.
Wireless connectivity capabilities are also very complete: the BoltGPU 10-31 provides Wi-Fi 6 and Bluetooth 5.1; an optional LTE cellular adds even more flexibility, while a multi-constellation GNSS with Untethered Dead Reckoning provides geolocation (factory option: high precision RTK GNSS).
The unit features rugged M12 connectors and is EN50155, EN45545 and E-Mark certified; it also has a wide range power input (24VDC to 110VDC, EN50155) with ignition sense and EN50155 OT3 operating temperature range (-25°C to +70°C) so it is suitable for rolling stock, automotive and heavy-duty applications. The BoltGPU 10-31 shares the same form factor of the BoltGATE 20-31 and of its expansion modules, making it possible to create hybrid architectures with extra features.
Aaeon | www.aaeon.com
ADLINK Technology | www.adlinktech.com
Aetina | www.aetina.com
Axiomtek | https://us.axiomtek.com
Cincoze | www.cincoze.com
DFI | https://us.dfi.com
Eurotech | www.eurotech.com
Lanner | www.lannerinc.com
Nexcom | www.nexcom.com
OnLogic | www.onlogic.com
Sintrones | www.sintrones.com
Vecow | www.vecow.com
VIA Technologies | www.viatech.com
PUBLISHED IN CIRCUIT CELLAR MAGAZINE • JANUARY 2022 #378 – Get a PDF of the issueSponsor this Article
Jeff served as Editor-in-Chief for both LinuxGizmos.com and its sister publication, Circuit Cellar magazine 6/2017—3/2022. In nearly three decades of covering the embedded electronics and computing industry, Jeff has also held senior editorial positions at EE Times, Computer Design, Electronic Design, Embedded Systems Development, and COTS Journal. His knowledge spans a broad range of electronics and computing topics, including CPUs, MCUs, memory, storage, graphics, power supplies, software development, and real-time OSes.