All products were visually inspected by inspectors. The severe inspection standards placed a heavy burden on the inspectors.
In addition, the inspectors had to overkill in order to avoid variations and defective product leakage.
For die-cast products, it is difficult to develop a rule-based inspection algorithm to detect defective products by creating rules for each defect to be detected. Therefore, it was thought to be difficult to automate inspection.
Also, in the past, it was thought that automation of conveyance such as visual inspection and picking would cost tens of millions of yen. In other words, not only the technical aspect but also the return on investment was an issue.
By utilizing the robot control and open source software provided by our company, we made it possible to automate picking, imaging, and visual inspection from a state of bulk. In addition, we realized automatic picking using commercially available cameras and distance sensors. By combining our self-developed AI software with collaborative robot arms and lighting, we have realized an automatic inspection system at low cost.
This device realizes automation of inspection and performs big data analysis of defect patterns. The results of this analysis can be fed back to the casting and machining processes to improve the manufacturing process.