We have a UR3 that is occasionally missing a waypoint and not by a small amount.  This issue popped up on a UR5 earlier this year and we thought it was a isolated issue, this robot was actually resolved by one of the software updates either 3.2.20 or 3.2.21, I don't recall which.  So this robot when it picks up the part out of the chute its missing occasionally by as much as 6-8 mm from the actual waypoint.  We are trying to catch a video of it doing this but its not very often that it occurs.  When it does occur sometimes it will happen several times in a row and then go right back to working like normal.  This is a robot that processes about 2000 parts a day and we may only see this a couple times a day.

The UR3 is running 3.3.0 software.

Anyone else seeing any of these issues?

@matthewd92  You may have already looked at this but, is the waypoint blended?  I've seen where the controller will skip waypoints if blends overlap.  Should write a message to the log saying this occurred.  If it is only periodically it could be just on the hairy edge and every once in a while the math works out to skip

@Grady_Turner we do not have any blends in this area due to needing to hit the holes precisely.  Neither time that we have seen issues like this have we had any.  Its like the robot occasionally comes up with a different actual position of the arm.  It will miss several times in a row and then go right back to running normal and may run normal for an hour or a week.

@Stefan_Stubgaard  Can you offer any input here?

Good practice is always to update to latest software, which currently is 3.3.2

A wild shot: can it be that one or several waypoints are located inside a conditional statement like IF, and then occasionally the waypoints are skipped due to the condition? If a condition is a sensor input, it can potentially flicker and cause such behavior.

Not inside of an if. They are approach points for picking up the part so one point is about 7-8 mm above the part, the next point is a Z move in the tool direction straight down from the approach point. These are in a movel with a stop at each of the points. The second point is not relative to the first point. 

@Stefan_Stubgaard We we have not upgraded to the latest as we are 9 hours away from the robot and honestly until we have time to test the new firmware we do not install on customer robots as we have had issues in the past introducing new bugs into systems that were running fine. 

@matthewd92  It is difficult to tell what can cause this behavior.

If possible, you should sent program and installation file including video/pics of the application to the tech support in US for examination.