Accuracy versus Repeatability

Whenever you see a data sheet for an industrial robot you will see that a figure is quoted for repeatability. This was not always the case. In the early days there was much confusion regarding accuracy and repeatability.

The industrial robot is inherently an extremely repeatable machine. Teach it a point and it will return to it unerringly. This ability should not be confused with accuracy which implies that if you program in a location as a set of co-ordinates, then the robot will achieve that location in space, relative to it's base.

The problem occurs because not only does the computer have to calculate the points on a path for 6 axes in real time, but it has to do this in the case of a revolute arm configuration, for 6 'rotating' axes.

Manufacturing tolerances and long levers combine with the resolution of encoders to magnify the smallest inaccuracy to give a small, yet sometimes significant deviation from the required position.

Over the years various techniques have been applied to try to give the robot the same degree of accuracy as say a co-ordinate measuring machine, and they have not been without success. A technique was developed for each arm to determine it's own arm signature by error mapping and recording any unique deviations.

Arm signature was a success but it soon became clear that high repeatability was the only attribute required to achieve success in 98% of robot applications. It is no surprise that high accuracy equipment like CMM and 3D routers favour 3 linear major axes to achieve their goals.