Object identification and 3-D position calculation using eye-in-hand single camera for robot gripper

Abstract

In this paper we designed an intelligent gripper module with vision system for robot manipulators. The vision system in the gripper module used only a single camera, as in the form of eye-in-hand architecture, for identifying a target object in space for the robot manipulator to grasp. SURF algorithm and back projection method of hue-saturation histogram model are applied for correctly identifying the target object. Then the identified object position in space can be located by employing the binocular vision system model using images captured at two different positions. Experimental results show that the vision system can correctly identify the target among different objects and calculate its position with relative error less than 4%. Finally, the intelligent gripper module is attached to a robot manipulator and experiments are carried out to verify its effectiveness in locating objects in space for robot arm to grasp successfully.