when I first started working with vision systems. I am still pretty junior mind you. I was told the 4 most important things for vision are.
1. Lighting
2. Lighitng
3. Lighting
4. Contrast.
My company integrates Vision Systems as one part of our business. We favor these two but we're recently impressed by what Balluff had to offer. Their SmartCamera line is very impressive and runs on their BVS-****pit which is entirely web based. I did a week long training class up at their facility in Kentucky. One of the best vision courses I have attended.
Whenever we start a new program I always ask "will this need Vision" 9 times out of 10 they tell me no and then end up coming back and asking "can we put a camera on this?" To which I always reply "What do you mean by we?" but my point is its a lot easier to design a inspection from inception rather then trying to back engineering it.
Lighting is very important but if you don't present the parts consistantly you'll still have problems. If your doing injection insert molding and picking the parts with a robot you need a good EOAT so that it picks the parts so that they are located precisely when you present them to the camera every time. Otherwise your going to have a high failure fallout rate or possibly pass bad parts. Im sure your aware in the auto industry even one part in a million they will put you on a sort and if they find more they escalate and it becomes very expensive. It might not be possible to see the defect your looking for in the EOAT so you might have to fixture them and even automate the fixture to see the defect. The problem with all this is it adds cycle time. If your molding 16 parts every 10 seconds and the inspection adds 2 seconds to the cycle time thats usually not acceptable.
The other thing I see a lot is once you put a camera on a machine they tend to rely on detecting defects rather then molding good parts.
