Active uframe or robot "world" being incorrectly set would not cause the robot to move on a curved way if doing cartesian movements, it only would cause to move on an incorrect direction, but the movement should be still linear, even if it goes in a diagonal direction.
If you are experiencing a "J shaped movement", I have given you two possible causes, forget about uframes, utools or cell frames until you can jog the robot lineary without strange motions.... if you jog the robot in world, user or tool it still should move lineary even if not in the correct direction, robot moving on a curved pattern or "falling" middle linear movement is not normal.
Regarding power loss, it is not a problem if the robot batteries that keep the encoders with power still have charge, these bateries are installed on the base of the robot and should be changed from time to time WHILE the robot is still powered on. These batteries are normal non-rechargable batteries that you can purchase on any shop.
If batteries got empty you should have received a "BLAL" alarm (Batery low alarm) or even a "BZAL" alarm (Battery zero alarm).
If you do not change the batteries after a BLAL you run the risk of losing mastering completelly if a power outage occours.
If this happens, the only way to fix it if to remaster each axis manually again with the zero marks, but it's very difficult to achieve the same precission on the robot as from factory when you do it manually... and you also say that robot has vision, that is the worst case scenario since a robot with vision needs to be very well calibrated in order to work correctly.
I repeat, have you checked if robot goes to 0 position AFTER mastering the axis again??? To check it the better way would be to create a new program, record a new point, change it to JOINT representation with fine termination, modify all axis values to 0 degrees and executing the motion.
After doing that you should check if robot moved to zero marks, if not, master the robot again.
Lastly, I don't know if I understood you correctly with this statement:
Quote
The imaged restored is from the same machine made earlier this year
If the image came from the same exact CONTROLLER, so you made a image backup from that controller and you restored his own image on his own controller, then mastering data was likelly lost since it should restore the original mastering after restoring the original image. What loses the pulse count after a power loss are the pulsecoders, so even if mastering is ok on the controller side, new pulse values will be generated on the encoder (robot) side so mastering data is invalid after a complete power loss on the encoders)
If instead you are saying that you restored an image from a similar robot (even if same mechanical unit) from another cell that was installed this year into this robot that is from 2013, this is probably the problem.
You should never restore a image from a controller into another, as all options, data and even mastering from the robot you extracted the image will be restored, and mechanical units all have different mastering data, even if they are exactly the same model.