Eye-to-hand coordination: a coarse-calibrated approach

摘要

This paper integrates different novel concepts to perform eye-to-hand coordination and manipulation to realise a working robot named COERSU. One of the salient features of our system is that it is not highly dependent to the camera-robot calibrations. Our coarse-calibrated eye-to-hand visual servoing uses monocular vision and primitive tactile information to extract 3D information of the scene. A robust genetic tuner is briefly presented to optimise the parameters of the image segmentation and edge detection. Some simple methods for scene classification, pose detection and final verification were also implemented to perform intelligent grasping. Experimental results from COERSU in a table-top scenario to manipulate some soft objects (fruit and egg) are also provided to validate these methods (snapshots and videoclip). 1