Advantech, an edge computing solution company, has recently unveiled a range of new industrial AI platforms based on Qualcomm Dragonwing IQ-9075 processor. Designed for industrial vision AI, the new product range comprises of a AOM-6741 SMARC computer-on-module, ASR-A503 and AFE-A503 robotic controllers, and an AIR-055 edge AI system, are the result of Advantech’s continuous research and development efforts to bring high performance and low latency to AI at the industrial edge.
With a performance of 100 TOPS (dense) and 200 TOPS (sparse) for AI computation, these systems support 16 camera feeds in parallel, and enable visual intelligence at industrial level. This means, from regular machine visions to advanced real-time visual reasoning and multi-stream analytics, the industrial edge can handle the processing load very efficiently.
High-Throughput Processing for Multi-Camera Vision Intelligence
Vision intelligence for multi-camera applications in high-throughput environments A Qualcomm Dragonwing IQ-9075 processor is at the heart of the system, and has an image signal processor (ISP), video processing unit (VPU), and neural processing unit (NPU) as its special features. Thanks to high-speed connectivity interfaces like MIPI-CSI, USB 3.0/Gigabit Ethernet, the system can carry out low-latency image preprocessing, multi-stream video decoding, and large-scale camera deployments.
Also Read: Kioxia and Sandisk Partners to Unveil 9th-Gen 2Tb QLC 3D Flash Memory to Power Enterprise AI Infrastructure
Heterogenous computing engine of the processor can solve computer vision problems like automated optical inspection, optical character recognition (OCR), real-time object tracking, etc. keeping the system under thermal and power limits.
Core technical highlights of the hardware architecture include:
Heterogeneous Processing Unit: Combines an 8-core Qualcomm® Kryo™ Gen 6 CPU with an integrated real-time microcontroller unit (MCU) subsystem to ensure deterministic performance for safety-critical and latency-sensitive industrial applications.
Scalable Sensor Ingestion: Native hardware support for up to 16 camera inputs, enabling broad spatial coverage across automated manufacturing cells and autonomous mobile robots (AMRs).
High Efficiency-per-Watt: Optimized Arm-based computing architecture designed to maintain high AI inference throughput within fanless and space-constrained industrial enclosures.
Streamlined Software Infrastructure and Edge Deployment
To reduce engineering cycles and speed up time-to-market, Advantech pairs the hardware lineup with a modular software integration layer. The company’s proprietary BSP Launcher simplifies board support package management and OS image deployment.
Additionally, the Advantech Edge AI SDK and Robotics Suite provide pre-integrated artificial intelligence frameworks, optimized multimedia pipelines, and Robot Operating System (ROS2) developer tools. This unified software ecosystem allows system integrators to rapidly deploy, manage, and scale vision reasoning workloads across modern factory environments.
