Vecow RAC-1000 rugged Edge AI systems feature NVIDIA Jetson AGX Orin, waterproof ports, GMSL2 camera connectors

Vecow’s RAC-1000 series Edge AI systems are powered by NVIDIA Jetson AGX Orin 32GB or 64GB system-on-modules, offering up to 275 TOPS of AI performance. These systems are energy-efficient and come with rugged I/O options and an IP67-rated enclosure, making them suitable for AI and robotics applications such as automated agricultural machinery, construction automation, and mobile robotics in extreme outdoor conditions.

The series includes two models: the RAC-1000, which supports 8 GMSL2 cameras through FAKRA-Z connectors for autonomous mobile robots, agricultural vehicles, and ADAS; and the RAC-1100, which features 6 GigE LAN ports with 4 PoE+ for vision AI applications. Both models are built for industrial environments and support various AI-driven tasks.

As you may know, we’ve previously covered several Vecow AI computing systems, including EAC-5000, EVS-3000, TGS-1000 Series, SPC-9000 fanless embedded system, ECX-3200, and EPBC-1000. Feel free to explore these options if you’re interested.

The Vecow RAC-1000 series specifications:

• System-on-Module
  • RAC-1000 R32 and RAC-1100 R32 – NVIDIA Jetson AGX Orin 32GB with
    • CPU – 8-core Arm Cortex-A78AE v8.2 64-bit processor @ 2.2 GHz with 2MB L2 + 4MB L3 cache
    • GPU
      • NVIDIA Ampere architecture with 1792 NVIDIA CUDA cores and 56 Tensor Cores @ 1 GHz
      • DL Accelerator – 2x NVDLA Engines
    • Video Encode – 1x 4K @60 | 3x 4K @30 | 6x 1080p @ 60 (HEVC)
    • Video Decode – 1x 8K @30 | 7x 4K @30 | 11x 1080p @ 60 (HEVC)
    • System memory – 32GB LPDDR5 Dynamic Random Access Memory (DRAM)
    • Storage – 64GB eMMC 5.1 flash
  • RAC-1000 R64 and RAC-1100 R64 – NVIDIA Jetson AGX Orin 64GB with
    • CPU – 12-core Arm Cortex-A78AE v8.2 64-bit processor @ 2.2 GHz with 3MB L2 + 6MB L3 cache
    • GPU
      • NVIDIA Ampere architecture with 2048 NVIDIA CUDA cores and 64 Tensor Cores @ 1 GHz
      • DL Accelerator – 2x NVDLA Engines
    • Video Encode – 2x 4K @60 | 4x 4K @30 | 8x 1080p @ 60 (HEVC)
    • Video Decode – 1x 8K @30 | 7x 4K @30 | 11x 1080p @ 60 (HEVC)
    • System memory – 32GB LPDDR5 Dynamic Random Access Memory (DRAM)
    • Storage – 64GB eMMC 5.1 flash
• Storage – 2x M.2 Key M 2280 sockets
• Video
  • 1x Digital Display interface (DDI) up to 4K60
  • 1x HDMI via M12 connector
• Camera – RAC-1000 – 8x Fakra-Z connectors for GMSL 1/2 automotive cameras
• Connectivity
  • 2x 10/100/1000 Base-T Ethernet GigE LAN with X-coded M12 Connector
  • RAC-1100 additional LAN – 4x IEEE 802.3at (25.5W/48V) GigE PoE+ LAN with X-coded M12 Connector
• USB – 2x USB 3.1 ports
• Serial
  • 2x COM RS-232/422/485 with M12 A-coded connector
  • 2x isolated CAN Bus with CAN FD support (M12 A-coded)
• Expansion
  • M.2 Key B socket (3042/3052)
  • M.2 Key E socket (2230)
  • PCIe x8 slot
• Misc
  • Power Button with LED and Recovery button
  • Micro USB port for OS flashing
• Power
  • Input – DC 9V to 50V via 4-pin K-Code M12 connector
  • Ignition Control – 16-mode Software Ignition Control
• Dimensions – 330 x 260 x 80 mm
• Weight – 7 kg
• Temperature Range
  • Operating temperature
    • 30W TDP Mode: -25°C to 70°C, with 0.63 m/s airflow
    • 50W TDP Mode: -25°C to 55°C, with 0.63 m/s airflow
  • Storage temperature: -40°C to 85°C
• Humidity – 5% to 95% Humidity, non-condensing; 95% @ 70°C relative humidity
• Shock – Operating, MIL-STD-810G, Method 516.7, Procedure I
• Vibration – Operating, MIL-STD-810G, Method 514.6, Procedure I, Category 4
• EMC – CE, FCC, EN50155, EN50121-3-2, IP67

The RAC-1000 series supports the NVIDIA JetPack 6.0 SDK based on Linux for the deployment of AI workloads. The systems support wall-mounting installations.

As noted in the specifications, the NVIDIA Jetson AGX Orin-based RAC-1000 Edge AI systems are available in four variants: RAC-1000 R32/R64 and RAC-1100 R32/R64 with different ports and module options. However, the company hasn’t provided availability or pricing details yet. We can expect this type of hardware to cost a few thousand dollars. More information about the four variants can be found on the product page.

Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress