Sintrones IBOX-601-G2-ONX16 NVIDIA Jetson Orin NX 16GB In-Vehicle Edge AI Computer with 4x GMSL-2 Cameras

Multi-camera perception in a vehicle tends to grow: more cameras get added, models get larger, and memory that looked generous at the pilot stage fills up by fleet rollout. The IBOX-601-G2-ONX16 is the build of the Sintrones IBOX-601-G2 series specified for that path, an in-vehicle edge AI computer that pairs the NVIDIA Jetson Orin NX 16GB module with four GMSL-2 automotive camera inputs.

The Orin NX 16GB combines a 1024-core NVIDIA Ampere GPU with 32 Tensor Cores, an 8-core Arm Cortex-A78AE CPU, two NVDLA v2 deep learning accelerators and 16 GB of LPDDR5. Against the IBOX-601-G2-ONX8, that doubles the memory and the accelerator count, leaving room for larger detection models and several camera pipelines resident at once. The IBOX-601-G2-ONX16 ships with NVIDIA JetPack, so existing Jetson Linux vision stacks carry across.

Four GMSL-2 inputs on FAKRA-Z connectors take automotive cameras over locking coaxial runs. Networking is one 2.5GbE and one GbE port on M12 X-coded connectors, alongside CAN FD and RS-232/422/485 on DB-9, two USB 3.2 ports sharing 10 Gbps and a 4K-capable HDMI output. An M.2 2280 NVMe SSD holds the system image and recordings, while M.2 Key B with a nano SIM slot, M.2 Key E and full-size mPCIe take optional 4G or 5G, Wi-Fi and GPS modules.

Power arrives on a 9-60 V DC M12 K-coded input with smart power and ignition management, so the system follows the vehicle start and stop cycle. The in-vehicle edge AI computer runs fanless from -25 °C to 70 °C, is tested to MIL-STD-810G shock and vibration and IEC 60068-2-64, and carries E-Mark with EN 50155 and EN 45545-2 (R25) compliance. Sintrones supports DIN rail and wall mounting for the IBOX-601-G2-ONX16.