A Differential Evolution Approach for Camera Space Manipulation

This paper proposes a Differential Evolution (DE) method for Camera Space Manipulation (DECSM). Development of DECSM was inspired by the realization that Camera Space Manipulation (CSM) is sensitive to errors in the geometric characterization. In order to increase the precision of robot maneuvers, the coordinates of “cues” (several points on the end effector) must be more accurate. This need is due to the fact that any amount of error in the designated cue positions relative to the tool juncture adds the same amount of error to the final three-dimensional placement. DECSM tests various combinations of adjustment of these offsets in order to optimize the model using extremely large amounts of data.