Advantech AIR-120 Brings Vision AI to the Edge

Advantech has introduced the AIR-120, a new compact edge AI system designed to accelerate real-world AI deployment in industrial and embedded environments.

At its core, the AIR-120 is built to run vision-based AI tasks efficiently at the edge, combining a choice of Intel Atom™ x7433RE, Core™ i3-N305, or Processor N97 CPUs with the Hailo-8 AI accelerator. Together, this platform delivers up to 26 TOPS of inference performance, enabling accurate, low-latency processing without relying on cloud resources. It’s an ideal setup for scenarios like public safety monitoring, AI-assisted inspections, and worker safety applications.

Designed with reliability in mind, the AIR-120 can operate in harsh conditions, with a wide thermal envelope from -20°C to 60°C. EMC compliance with IEC 61000-6-4 and 61000-6-2 ensures stability in electrically noisy industrial environments. The compact housing (176 x 112.35 x 66.4 mm) includes a dedicated thermal solution to maintain Hailo-8 performance even under sustained load, such as multi-stream object detection using YOLOv3.

Rich I/O and Modular Expansion

For system integrators, I/O flexibility is critical. The AIR-120 offers:

• Dual LAN (2.5GbE and 1GbE)
• 2 x USB 3.2, 2 x USB 2.0
• 2 x COM ports, optional dual CAN bus
• 3 x M.2 slots for Hailo-8, storage, Wi-Fi, or 4G/LTE modules

One Ethernet port supports EdgeBMC for secure, out-of-band remote management, making the unit easier to deploy and maintain at scale.

Software Tools for Seamless AI Development

The system ships with Advantech’s Edge AI SDK and Hailo’s software stack, including:

• Model Zoo – Pre-trained, ready-to-deploy models
• Dataflow Compiler – For efficient model conversion
• TAPPAS Toolkit – Examples for object detection, segmentation, and pose estimation
• Benchmarking tools – For evaluating performance across real workloads

Combined with remote management via EdgeBMC, the AIR-120’s software ecosystem simplifies the entire AI deployment process, from model development and benchmarking to optimisation and monitoring. This reduces the complexity typically associated with edge AI projects and shortens time to value.

With its compact footprint, industrial-grade reliability, and strong developer support, the AIR-120 is a ready-made platform for deploying vision-based AI at the edge. It’s a practical choice for teams looking to integrate AI into safety, inspection, or automation applications without overhauling existing infrastructure.

Read the original announcement here.