Advanced Automated Piston Head Inspection

Impetus to Invent

Human operators are, well, human, and when your product demands perfection, machine vision can answer the call! People are extremely good at picking out details, and it’s not hard for an inspector to spin a part in their hand and look for defects. So how do you automate this process? Invent Automation to the rescue. By incorporating 10 cameras, 11 images, and a few single-axis robots, we run the part through the machine vision gauntlet using a variety of techniques and specialty lenses to make a judgment call on whether the piston is good or bad. From there, we automatically sort the product as good, bad, or suspect. In the case of a suspect part, an operator can further inspect it. Now, the operator is freed up to do other, more value-added activities, and the end customer can be assured that defects are a thing of the past.

Problem Solved

People aren’t perfect, but machine vision is and never slows down for a break.

Lessons Learned

All machine vision projects on high-volume lines suffer from the same pitfalls: defect samples can be hard to come by! Defect samples are a critical part of a machine vision project's success

What Was Unique

Combining 10 cameras, including a line scan camera, and 3 single-axis robots, all within an extremely tight footprint on an existing line, was an exciting experience!

What Was Challenging

The whole machine had to be integrated into about an 18-inch-wide space at the end of a line. Making it wider would mean crossing into the aisle way, which was forbidden. Luckily, we excel at difficult tasks and had no problems squeezing so much machine vision inspection into such a tiny space.