Forecr DSBOX-ORNX-AC8 NVIDIA Jetson Orin Nano 8GB Industrial-Grade Embedded Computer

Many edge AI installs run one or two neural networks against a camera feed, and the deciding factor is how much memory the model set needs. For that middle ground the DSBOX-ORNX-AC8 fits the NVIDIA Jetson Orin Nano 8GB into a compact industrial-grade embedded computer from the Forecr DSBOX-ORNX series, sized for machines, robots and control cabinets.

The Jetson Orin Nano 8GB module combines an NVIDIA Ampere GPU with Arm64 CPU cores and 8 GB of 128-bit LPDDR5, twice the memory of the 4GB tier. That headroom lets the DSBOX-ORNX-AC8 hold a larger detection network, or a second model alongside the first, without paging. The computer supports Ubuntu Linux 20.04 with JetPack 5.1, so software developed on a Jetson Orin developer kit moves across with little change.

For I/O the DSBOX-ORNX-AC8 provides one Gigabit Ethernet port, two USB 3.1 Type-A ports, an HDMI 2.0 output at up to 3840 x 2160, one CAN bus port, one RS232 and one RS422 port, and two digital inputs with two digital outputs, plus a micro USB 2.0 port for recovery. Two M.2 Key-M slots take SSD storage, one 2280 socket on four PCIe lanes and one 2230 socket, while an M.2 Key-B socket with SIM slot and an M.2 Key-E socket add LTE, 5G or WiFi modules, so a remote unit can report over cellular without an external router. Power comes from a 9-28 VDC input that suits common 12 V and 24 V supplies.

Thermal management cools the module without a fan across a -25°C to +85°C ambient range, inside a 110 x 130 x 60 mm embedded computer weighing 760 g. Forecr names robotics, transport and logistics, smart city, energy and agriculture among its target applications. Where a lighter model set fits in 4 GB, the DSBOX-ORNX-AC4 uses the Orin Nano 4GB, and the DSBOX-ORNX-AB8 moves to the Jetson Orin NX 8GB for heavier inference.