Evaluation of Hand-Eye Calibration Algorithms in Application to Robotic Flexible Manufacturing Systems.

The article is focused on two vital practical aspects related to robotic flexible manufacturing system which can be used for example in small craft factory to support the production of short series: the automatic visual system calibration with respect to the workspace and visual work-piece localization. The test procedure was performed in the system consisted of a collaborative robot UR-10, HD RGB camera and 8-cameras Optitrack motion capture system used as a reference. For calibration three hand-eye algorithms were investigated, where the best accuracy enabling to localize workspace coordinate system was achieved for dual-quaternions approach with the error below 0.4 mm in each direction. The work-piece localization was performed by means of \(40\times 40\) mm ArUco tag, the mean localization error was below 0.6mm in every dimension.