Experiments to achieve accurate vision guided robot movements

Highly accurate sensor based robot movements are demanded in many light assembly applications. In our study the goal is to implement a vision sensor in the gripper frame of a robotic manipulator to investigate how to achieve high accurate but at the same time flexible movements. Our sensor comprises two cameras and a suitable lighting. The 3D coordinates of the objects close to the grasping point are extracted based on the normal stereo disparity calculations. The recent results of our experiments shows that we are able to achieve the accuracy of about 20 micrometers in the area of 90/spl times/70 mm/sup 2/ and that our control method is capable of driving the robotic manipulator according to this accuracy. In this paper we describe our control method and the performance of the sensor as well as the preliminary results of our studies.

[1]  Peter I. Corke,et al.  A tutorial on visual servo control , 1996, IEEE Trans. Robotics Autom..

[2]  Janne Heikkilä,et al.  A four-step camera calibration procedure with implicit image correction , 1997, Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[3]  Peter Corke,et al.  VISUAL CONTROL OF ROBOT MANIPULATORS – A REVIEW , 1993 .

[4]  Gregory D. Hager,et al.  A modular system for robust positioning using feedback from stereo vision , 1997, IEEE Trans. Robotics Autom..