The repeatability of the UR is listed as 0.1 mm.  While this may be true for point-to-point positioning, in my experience the path between those points will not be as precise.  We've encountered several path-critical applications where the UR has fallen short--in one example, while traveling in a straight line between two waypoints with the moveL command, the position of TCP would fluctuate up and down by up to 2mm along the supposedly linear path. This occurs even in tests with no payload.

In your experience, does the use of path teaching with the FTS improve the repeatability along a path? Are there any other ways to achieve a highly repeatable path?

@jbahner there are a few ongoing discussions linked to this.

This one is about UR movements preicision: http://dof.robotiq.com/discussion/312/ur-path-problems-on-small-radii

This one is about our Path recording precision: http://dof.robotiq.com/discussion/313/vision-locate-together-with-path-recording

Thanks @Etienne_Samson those links are both helpful!  Please see my follow-up response to a comment you made in the "Path recording precision" thread (link below).  I've had some recent experiences that suggest the +/-0.1mm repeatablity spec listed by UR does not hold true while traversing an interpolated path between two programmed waypoints.  I'm interested to know if the FTS can improve the path precision, or if it is limited by the path precision of the UR.  

http://dof.robotiq.com/discussion/313/vision-locate-together-with-path-recording

@jbahner your welcome, I saw your comment and I'll respond overthere, you got some good points. I understand know what you want to do and why you care about the Path precision.