IRVision problem with stability

    Hello,


    I have problem with stability of the vision process in bin picking application (from drawer). There are two robots (R30iB) in the production line, which are the same job and nearly the same software and configuration (Sony camera, red LED light near the camera). Main diference between them is in the sizes and shapes of components in the drawer (the have diferent tool sets).


    One of them works properly, but second robot make a troubles. This robot works also properly for few months, but now seems, that vision process results are wrong (only sometimes and only for some types of components). Robot works with aprox. 8 types of component (it has tool changer and different tool for each of it). The problem is only for two of them (it is strange). For the operators on the line it seems, that the robot is trying to get component 1mm next to correct position (in X direction). So they have to correct reference position for vision process and it works for next aprox 5-10 pieces. After that it is needed to make 1mm correction again and again to ensure continual process. After 3-5 corrections it seems, that it is needed to make correction in oposite direction - they uses original reference position and it works. This problem appears irregulary and repeats several in a day.


    We checked mechanical things (tool stability, camera holder and also LED light) and we can exclude it. We also checked the light behavior in the production hall. It is without windows (only artificial lighting) and nothing has changed in the last few month. It seems, that the vision process works well, the resulting score is usualy greather that 80%. There are not visible shadows or reflections in the runtime view, so the resulting offsets shold be OK.


    Now I am clueless. Did You meet someone a similar problem? I would appreciate any advice.

    Have you excluded the possibility of a mechanical or encoder problem? Could the robot have crashed? Maybe you can use a bit of oil based clay and have the robot poke one spot each cycle, you could see over time if its hitting the same spot.

    Thank You for Your idea. I checked (remotely in cooperation with customer) all possible sources of problems. My personal intervention was little complicated, because the customer is too far from me now.


    But fortunately the problem is solved, case is closed. The problem was in the cable between camera and robot. Realtime image from camera became worse and was not stable. Last week it contained even visible shaking stripes. From the beginning, the problem showed irregularly. But last week the image was unusable and problem became clear.

Advertising from our partners