Axiomtek
Facebook Twitter LinkedIn YouTube Instagram

Automated Optical Inspection Changes the Game for Manufacturing

Automated optical inspection is a turning point for manufacturing; it has been widely utilized in industries such as PCB and semiconductors. It improves yield rate and efficiency with the assistance of AI algorithms to spot defects, becoming an essential part of smart manufacturing.

Challenges

The customer, a wafer foundry, was seeking a hardware and software integration partner to deploy AOI for wafer production line inspection. In addition to a high-performance computing platform, the customer required a toolkit to optimize deep learning models and simplify deployment on Intel platforms.

Main Requirements

  • High-performance industrial platform
  • GPU to accelerate visual computing
  • High-resolution industrial camera for capturing images
  • Software toolkit to train and run deep learning models for different inference tasks

The Fanless Smart Edge Computing System Provides Exceptional Performance with Flexible Configurations

Axiomtek recommends the industrial system IPC950 to run AOI. The embedded platform features the 11th gen Intel® Core™ processor (Tiger Lake) with Intel® Iris® Xe graphics for high-performance graphics and AI inference. It supports a PCIe x4 slot for accelerator cards, operates in -20°C to +50°C, and includes a full-size PCIe Mini Card slot, SIM card slot for Wi-Fi/Bluetooth/LTE, and M.2 Key E 2230 socket for wireless connection. Storage options include a 2.5" HDD/SSD bay and an M.2 Key B+M 2280 slot. Two DDR4-3200 SO-DIMM slots support up to 64GB of memory. Front I/O includes four USB 3.1 Type-A, one 2.5G LAN and two GbE LAN, one RS-232/422/485 COM port, and a programmable 8 CH DIO. Users can expand I/O via COM or LAN modules.

Improving Defect Detection in Wafer Fabrication with AI at the Edge

Improving Defect Detection in Wafer Fabrication with AI at the Edge

AOI defect detection starts with training data, consisting of thousands of images for deep learning. AI algorithms improve continuously to detect various defect types. Moving image processing, defect detection, and classification to the edge reduces data transmission and power consumption. A single system can detect different defect types in real-time, and with additional training, the visual detection algorithms can expand to spot even more defects. Automating defect detection ensures faster inspections, consistent accuracy, and eliminates human fatigue and errors.

The Axiomtek AI Suite (AIS) is an application-ready package based on TensorFlow, Intel® Edge Software Hub for Industrial, and OpenVINO toolkit. AIS simplifies the integration of advanced vision applications with automation, improving defect detection efficiency and accuracy.

System Configurations of the IPC950

  • Intel® Core™ i7-1185G7E with Intel® Iris® Xe (Tiger Lake)
  • M.2 Key E 2230 slot for Wi-Fi/LTE
  • Supports TPM 2.0
  • Supports Intel® iAMT
  • Wide operating temperature from -20°C to +50°C
 
 
Copyright 2024 Axiomtek Co., Ltd. All Rights Reserved