Beschreibung
Geometric calibration is well known to be an efficient way to enhance the absolute accuracy of serial as well as parallel robotic systems. While most calibration techniques require particular equipment, a sophisticated and advanced procedure for kinematic calibration of parallel manipulators has recently been reported which solely relies on the sensors of the robot”™s motors and hence shows an economic edge over traditional parameter identification approaches. Furthermore the procedure is completely automatable, thus belonging to the class of self-calibration methods. The strategy relies on passing type two singularities (i.e., direct kinematic singularities) and is therefore referred to as singularity-based calibration. One drawback of the method is, however, that redundant information necessary for parameter identification is just collected in singular configurations which amount only to a small part of the overall workspace. In order to make the calibration result valid within the whole workspace and as reliable as possible, though, it is important to carefully choose the measurement configurations utilized within the parameter identification step. This paper addresses this problem by adopting the concept of optimal pose selection for conventional calibration schemes to the singularity based calibration approach. Results are exemplarily presented for a simple planar parallel structure.