AI Vision Systems

Artificial Intelligence (AI) combined with machine vision is a powerful combination for applications with unpredictable variations. Traditional vision tools become impractical because they require predicting all variations and anomalies before they exist. AI vision tools are designed to solve these problems, enabling new ways to utilize machine vision.  

For instance, identifying scratches on parts, discoloration from annealing processes, poor quality barcodes, classifying parts that look almost identical to the human eye, and detecting defects in fabrics were applications that traditional vision tools couldn't easily or at all manage. 

---

CASE STUDY EXAMPLE: AI Vision System for PCB Recycling

PROBLEM

Recycling precious metals from discarded PCB boards was a complex problem for our customer to solve. It was not possible for humans to accurately and rapidly sort over 6,000 unique PCB boards into 20 different categories based on value to preserve these metals. Instead, a deep learning AI based vision application was required. 

SOLUTION

To de-risk this project, Arimation started by analyzing potential deep learning AI algorithms and systems. The solution had to deliver high throughput and accuracy, which required significant training of the model.  Arimation selected the Pekat Vision AI solution and initially built a proof of concept with two stations - a vision system station as well as a sorting station. Engineers were able to validate that the boards could be sorted effectively at the high rate of speed required.  The final solution included Basler camera hardware, Advantech industrial PC and Dorner conveyors. 

VALUE

With this solution, the customer will be able to maximize the monetization of their PCB board recycling service and ensure that these precious metals are reclaimed and preserved.