3D object pose estimation using stereo vision for object manipulation system

Abstract

Object pose estimation is one of the crucial parts in vision-based object manipulation system using standard industrial robot manipulator, particularly in pose estimation of the end effector of the robot arm to grasp the object targeted. This paper presents the utilization of stereo vision system to estimate the 3D (3 dimensional) object position and orientation to pick up and place the object targeted in an arbitrary location within the workspace. In order to accomplish this task, a calibrated stereo camera in the eye to hand configuration is used to capture the images of the object on the left and right camera. Then, the specific object feature is extracted and the 3D position and orientation of the object are calculated using image processing algorithm. Finally, the end effector of robot arm equipped with gripper will pick up the object targeted according to the object pose estimation output, and then place it to the desired location. The experimental results using 6 DOF robot arm are demonstrated and show the effectiveness of the proposed approach with good performance.