Aetina Corporation, an edge device manufacturer and AI solutions provider, has developed an advanced edge AI solution that upgrades the existing automated optical inspection system found in manufacturing plants. The goal: use edge AI to reduce manufacturing defects.
Leveraging Nvidia’s Metropolis for Factories will minimize the occurrence of inspected units being falsely identified as defective to less than 5 percent, the company claims. The system can also reduce the chance of misidentifying defect-free items as defective.
Aetina created its edge AI solution by utilizing its SuperEdge AI platform, which consists of a range of x86 AI training platforms that can be configured with Nvidia GPUs. Currently, one of the platforms is Nvidia-certified system 3.0, enabling the optimal performance of Nvidia’s AI software tools. Additionally, Aetina has future plans to introduce more AI training platforms that are compatible with the Nvidia software suite. This will accelerate the development and implementation process for various AI use cases.
What is an automated optical inspection system?
The manufacturing industry extensively relies on automated optical inspection system, which excels in carrying out accurate and precise inspections and measurements that surpass human capabilities. These systems integrate optics, mechanisms, and electronic control to replace human visual perception, while the software component mimics the human brain by processing signals received from the optical components.
With the advent of edge computing, systems process the incoming sensor data to process at the edge, reducing latency and improving performance. An automated optical inspection system benefits from this to increase its accuracy and minimize the error percentage of misidentifying defect-free items.
Despite the implementation of an automated optical inspection system, Aetina highlights a concerning 20 percent risk of misidentifying defect-free items as defective. This poses a challenge for factory workers, as it necessitates reinspection and consumes valuable time. However, Aetina asserts that their proposed edge AI-based system can significantly reduce the change of misclassification. This allows assembly workers to concentrate their efforts on rectifying genuinely defective items, improving overall efficiency.
Aetina creates an AI computing system with Nvidia IGX Orin platform
During the Computex 2023 event, Aetina also showcased its newly developed AI computing system, AIP-IGX-Series, which holds the Nvidia IGX Orin platform, purposefully tailored for applications within the industrial and medical sectors. With the capability to deliver up to 248 tera operations per second of AI performance, these AI systems are well-suited for industrial-grade scenarios that require higher levels of performance, durability, and security.
In response to the requirements of challenging industrial environments, Aetina has developed the AI computing series with different chassis and cooling designs. Users will also benefit from the advantages of a comprehensive software stack and well-documented resources that facilitate industrial and medical certifications.
Nvidia’s recently introduced IGX Orin platform is a non-Jetson edge AI development platform that offers the flexibility to add the Nvidia A6000 GPU add-on board for enhanced compute performance. Interestingly, this AI platform provides up to 248 TOPS of computing power, slightly lower than the AGX platform’s 275 TOPS. However, it compensates for this difference by offering additional computing performance support, such as the ConnectX-7 smart network interface card (SmartNIC) that enables a throughput of 200Gb/s.
Aetina plans to release the AIP-IGX series for sample testing in October, with mass production units becoming available in December of this year.
Aetina | edge AI | GPU | manufacturing | Nvidia | smart factory