NRU-110V

NRU-110V series is a low-power edge AI computing platform that supports eight automotive cameras. It is a turnkey solution that includes computing core, on-board GMSL deserializers, and a pre-installed board support package (BSP) to support selected cameras. Powered by NVIDIA® Jetson AGX Xavier™, NRU-110V offers 11 TFLOPS FP16 or 22 TOPS INT8 GPU computing with 30W power consumption. The power-efficient design not only extends operation time for battery-powered machines, such as autonomous mobile robots (AMR), but also reduces operating costs for mass deployed applications, such as intelligent V2X.

The video input of NRU-110V takes advantages of commercial off the shelf (COTS) automotive cameras, from water-proof ability, high dynamic range, auto white balance, to LED flickering mitigation. In other words, your AI applications will always obtain high-quality images regardless of lighting conditions, from rainy, bright sunny days, to pitch-black nights. NRU-110V supports multiple automotive camera sensors, including SONY ISX019, On Semi AR0147, and AR0231. With the unique onboard trigger system, NRU-110V can simultaneously capture images from eight GMSL cameras within millisecond channel-to-channel skew.

By incorporating 10G Ethernet, NRU-110V can acts as a camera sensor hub by streaming real-time images to another CPU/GPU powerful computer. In addition to the support of eight automotive cameras, it also has one isolated CAN bus for in-vehicle communication, one RS-232 for GPS input and isolated DIOs for connecting sensors/actuators. NRU-110V supports GPS pulse-per-second (PPS) input to support time synchronization among multiple platforms. For data storage, in addition to 32GB eMMC on the Xavier module, NRU-110V includes an M.2 2280 NVMe socket for fast SSD read/write performance. It also has one mini-PCIe socket for the WIFI or 4G module.

Combining its eight GMSL automotive camera support, significant TFLOPS for inference, power-efficient design, and 10GbE data transmission, NRU110V is ideal for AI-based vision applications that require continuous interaction with surroundings, such as AMR, ADAS, intelligent V2X, etc.