Just in time for the NVIDIA GTC, the AI developer conference in San Jose, California, Syslogic is adding four SPE (Single Pair Ethernet) interfaces to its RML A4AGX AI-enabled rugged computer. Combined with the NVIDIA Jetson AGX Orin 64GB SoM (System on Module), the rugged computer is the ideal hardware platform for sensor fusion and inference applications.
Syslogic, a member of the NVIDIA Partner Network, specialises in embedded systems for vehicles, mobile machines and robots. The new Rugged Computer RML A4AGX SPE features four single-pair Ethernet interfaces, also known as Automotive Ethernet or formerly BroadR-Reach technology. Unlike conventional Ethernet, which uses two or four pairs of wires to transmit data, SPE requires only one twisted pair of wires. This reduces the amount of wiring required. SPE also allows the connection of typical automotive sensors and enables data rates of up to 1 Gbps. Combined with the automotive Ethernet standard 1000Base-T1, the rugged computer is suitable for data-intensive applications such as streaming high-resolution surround cameras, GPS corrections or lidar point cloud processing.
On the Rugged Computer RML A4AGX, each SPE interface has its own dedicated network interface controller (NIC). This enables real-time communication with cameras and other sensors.
In a typical use case, the Syslogic Rugged Computer serves as an edge AI computer in a vehicle or mobile machine. It collects high-resolution video from multiple cameras, lidar point clouds and telemetry data via SPE, processes it and transmits the results over the same network.
SPE interfaces allow the AI rugged computer to plug and play with the electronics of modern vehicles or machines that have implemented ISOBUS. No external gateways are needed.
The SPE interfaces are complemented by eight GMSL2 camera inputs. This allows data from multiple high-resolution cameras and sensors such as radar or lidar to be acquired and processed simultaneously. The embedded computer can be connected to a vehicle’s bus system via CAN. A PPS input synchronises the unit with the sensors. As a further highlight, two integrated uBlox GNSS receivers enable real-time kinematic and heading functions without additional hardware. Vehicles can be located or positioned with centimeter-level accuracy. These features make the AI Rugged Computer RML A4AGX SPE suitable for sensor fusion applications such as autonomous or semi-autonomous driving.
The embedded computer is designed for use in vehicles and autonomous machines not only in terms of functionality, but also in terms of reliability, robustness and availability. It also meets the EMC requirements for certification to ISO 13766 and ISO 14892 for agricultural and off-highway applications.
The fanless, rugged computer is certified for a temperature range of –25 to +70 degrees Celsius and is IP67/IP69 rated. It is shock and vibration resistant and designed for 24/7 operation. Based on the latest generation NVIDIA Jetson SoM, developers can also be assured of long-term availability. NVIDIA guarantees availability until at least 2032.
The RML A4AGX is based on the Jetson AGX Orin™. The compact SoM (System on Module) features 275 TOPS of AI processing power and 64GB of RAM. The unique combination of a powerful CPU and GPU with the NIVIDIA Ampere architecture opens the door to a new generation of autonomous vehicles and machines.
The NVIDIA JetPack SDK is pre-installed for efficient application development. Developers benefit from a comprehensive development environment. The JetPack SDK includes a Linux kernel, Ubuntu desktop environment, and a complete set of GPU computing, multimedia, graphics, and computer vision libraries. It also supports DeepStream for streaming video analytics and ROS2 and Isaac Sim for robotics.
