Syslogic presented its latest AI (artificial intelligence) embedded system. With its Nvidia Jetson AGX Xavier SOM (system on module), the rugged computer RPC A3 masters demanding AI inference applications.
In recent months, Syslogics has expanded its portfolio of AI-enabled embedded systems with Nvidia technology. In addition to the previously presented AI rugged computer based on Jetson TX2 (CAN Newsletter reported), Syslogic is now launching the RPC A3 with Jetson-AGX-Xavier technology.
The version with Jetson TX2 SOM already allows sophisticated GPU-accelerated AI applications such as Intelligent Video Analytics (IVA). But the rugged computer based on Nvidia’s Jetson AGX Xavier, also serves a higher performance class.
The Xavier SOM features an eight-core CPU combined with a 512-core voltaic CPU. The product thus offers AI computing performance for functions such as the sub-steps for autonomous driving (self-driving loops). They range from sensor data fusion (perception) – the combination of various sensor data to form an overall environment – to diverse intermediate steps and the specification of driving behavior.
Pre-trained neuronal networks combined with GPU-accelerated data processing at the network perimeter enable autonomous decisions without human intervention and without connection to the cloud. This is not only interesting for autonomous driving, but also for robot applications or autonomous machines.
The company guarantees a high level of functional reliability due to the absence of moving parts, screw-on plug connectors, and an ultra-rugged housing. The housing complies with protection class IP67 and is thus resistant to dust, water, and chemicals. A Gore-Tex ventilation module ensures pressure equalization in the housing. The product is also suitable for the broadened temperature range from –25 °C to +80 °C at component level. For vehicle use, the computer has an integrated ignition controller and a disturbance-free CAN interface. It is also compatible with the CAN protocols J1939 and CANopen. Moreover, an E-Mark certification is in the preparation phase, said the company.
The device is used in railway and municipal vehicles, construction, agricultural machinery, automated guided vehicles, and in mining vehicles. The computer is also suitable for markets like Smart City, industrial automation, and traffic engineering.
It is is designed for an input voltage of 9 VDC to 30 VDC. Additionally, up to four Gbit LAN interfaces are available, which can be configured as PoE interfaces. All of the company’s AI embedded systems are delivered with a pre-installed Ubuntu board support package. Furthermore, a complete developer environment (Jetpack) including Cuda libraries is installed. The Nvidia software development kit (SDK) and a deep learning framework for implementation of applications are available for introduction to the AI world.
News and reports