A monocular-vision-based alignment technique is proposed to achieve accurate alignment of tool end in multi-station operation with an industrial robotic arm. To begin with

the system calibration is carried out to establish coordinates mapping between the concerning coordinate systems. After that

a set of cooperative targets is designed and arranged near the alignment objects on the workpiece. The expected poses of cooperative targets can be obtained by offline measurement and a task table for alignment targets is then built up. In the multi-station operation stage

driven by the task table

the current poses of the targetsare calculated by the vision system

and the deviations between the current poses and expected poses are calculated. Then parameters to move the robot arm are evaluated according to the deviations by the coordinates mapping. By iterating the above steps

the final alignment is worked out upon the designed threshold. The experimental results show that the alignment accuracy can reach 0.1mm in X and Y direction

0.2mm in Z direction

and the angle error is better than 0.1° in the measuring distance of 260mm.