Iterative Learning Based Controller for Motion Error Correction of an industrial Robot

This research project engaged in, dimensioning the effect of robot user programming, on the positional accuracy, delivered by an industrial robot. User programming was shown to effect the robot positional accuracy, by up to 0.4mm. The positional error was shown to vary within the robot work envelope. Introducing a correction would therefore need to be pose dependent. In a production setting, dimensioning the value of the correction for each pose within the workspace, is time consuming and becomes impractical. Therefore three novel algorithm's were created and used to learn the value of the correction variable, required for any given pose. The benefit of applying the algorithm's, includes, a standardized and automated approach, able to significantly reduce the effect of user programming on the robot positional accuracy.