Stereo Vision-Based Robot Servoing Control for Object Grasping

Abstract

In this paper, a stereo vision-based robot servoing control approach is presented for object grasping. Firstly, three-dimensional projective reconstruction with two free-standing CCD cameras and homogeneous transformation are used to specify the goal grasping position and orientation of a robot hand. Secondly, a stereo vision-based servoing problem is formulated, and a stereo vision-based servoing control algorithm which is independent of robotic dynamics is proposed. Using this algorithm, a set of velocity reference inputs can be obtained to control the motions and velocities of the robot hand during the visual servoing. Thirdly, the methods for coping with the time delay of image processing and the CCD camera calibration are put forward. Lastly, the effectiveness of the present approach is verified by carrying out several experiments on object grasping using a 6 degrees of freedom robot. Its stability and robustness as well as flexibility are also confirmed by the experimental results.