Multiple cameras-multiple target points visual servoing in large scale 3D manufacturing systems

Abstract

This article addresses the multiple cameras — multiple target points visual servoing system to correct the pose of robot end-effector in large scale manufacturing systems. The end-effector of a 6 Degree-of-Freedom (DOF) robot holds the workpiece and four cameras are used as visual sensors and installed on the workpiece to capture the image features of target points. Four pairs of light emitted diodes (LEDs) are installed above the workpiece configured in a way that, each pair LED can be observed by only one camera. One pair LED contains one target point for visual servoing and one complimentary point for estimating the depth of the corresponding target point to relevant camera. In this configuration, the depth of each target point to the corresponding camera is readily obtained by using the stereo vision method. Hence, the image Jacobian matrices can be precisely calculated during visual servoing. An imaged based visual servoing (IBVS) controller is designed to control the pose of the end-effector of robot and hence that of the work piece. By observing the four target points, the IBVS can drive the workpiece to the desired position in robot base frame. Extensive simulation has been carried out to verify the effectiveness of the proposed method.