Unleashing Qualcomm QCS6490: Accelerating Next-Gen Edge AI Solutions with AOM-5721

Advantech, Taipei, Taiwan, 08.09.2025 - Advantech (2395.TW), a leading provider of edge computing and edge AI solutions, is excited to announce the arrival of the AOM-5721, a SMARC-based Arm Computer-on-Module powered by the Qualcomm® Dragonwing™ QCS6490 System-on-Chip (SoC) family. Designed to deliver i3-class performance and up to 12 TOPS of AI inference with only 12W TDP, it offers the space efficiency and low power consumption required for demanding Edge AI applications in Automated Optical Inspection (AOI), industrial automation, and retail innovation.

Powerful AI Processing and Seamless Integration with Ready-to-Use SDKs

Advantech AOM-5721 features the Qualcomm® Kryo™ 670 CPU with eight Arm v8 cores running at up to 2.7GHz, delivering high-performance computing for edge AI workloads. Integrated ISP and VPU support 4K60 video decoding and pre-processing, making it ideal for imaging and visual inspection systems. These capabilities are critical for visual inspection, diagnostics, and multi-camera applications.

With dual GbE, USB 3.2, PCIe Gen3, MIPI-DSI, MIPI-CSI, and eDP, AOM-5721 ensures versatile peripheral integration and an optimized I/O for multi-camera workflows. Featuring low power consumption and high performance in a compact SMARC form factor, it enables flexible and efficient system designs. 

The AOM-5721 supports Yocto 4.0 and above, Windows 11 IoT Enterprise, Android and Ubuntu Linux, along with long-term driver maintenance and SDK updates. In addition, the platform supports the Qualcomm® SNPE (Snapdragon Neural Processing Engine) SDK for optimized AI processing, together with Advantech’s Edge AI SDK, which provides pre-validated AI models, development tools, and integration support. With these SDKs, developers can easily accelerate AIoT deployment in applications.

Versatile Edge AI Platform for Medical and Industrial Innovation 

Building on its processing power and flexible I/O, the AOM-5721 delivers targeted advantages for industries that demand real-time response, reliability, and a compact form factor. In the medical field, it supports portable ultrasound, blood analysis, and 3D dental modeling, enhancing accuracy and efficiency with the Qualcomm® SNPE SDK for optimized AI processing. Its Qualcomm® MIPI-CSI camera interfaces are particularly beneficial here, enabling advanced multi-view medical solutions.

In industrial automation, it excels in smart HMI and machine vision, enabling fast object detection, defect inspection, and automation. The four-camera support further enhances its capabilities for complex machine vision tasks requiring multiple camera inputs for defect inspection, facial and object recognition or surveillance. 

Smarter Retail Experiences with Advanced Camera and AI Capabilities

The AOM-5721 is not only for medical and industrial applications, but is also engineered to revolutionize retail experiences by offering advanced camera and AI capabilities at the network edge. Its integrated Qualcomm AI Engine features dual HVX (Hexagon Vector eXtensions) and 4K HMX (Hexagon Matrix eXtensions), enabling sophisticated on-device machine learning for expanding edge computing use cases.

For smart retail, the AOM-5721 supports up to four concurrent cameras with the Qualcomm Spectra™ ISP 570L, making it ideal for biometrics-based authentication and payments, as well as in-store robots for monitoring and inventory. This capability facilitates innovative applications like AI-powered in-store autonomous checkout and personalized recommendations, significantly enhancing the consumer journey and optimizing store operations. With Wi-Fi 6E offering multi-gigabit speeds and comprehensive OS support, the AOM-5721 ensures premium, security-rich connectivity and flexible design, driving digital transformation in retail environments with long-term longevity support.

Advantech's AOM-5721 SMARC module, powered by the Qualcomm QCS6490 System-on-Chip (SoC), combines powerful AI computing, excellent connectivity, and rich features, bringing new possibilities for edge computing and AI applications.

The module is expected to be available in Q4 2025.