Near-Minimum Time Visual Servoing of an Underactuated Robotic Manipulator

Abstract

Rate of convergence to the desired pose to grasp an object using visual information may be important in some applications, such as a pick and place routine in assembly where the time between two stops of the conveyor is very short. The visually guided robot is required to move fast if vision is to bring the sought benefits to industrial setups. In this paper, the three most famous techniques to visual servoing, mainly the image-based, position-based and hybrid visual servoing are evaluated in terms of their speed of convergence to the grasping pose in a pick and place task of a momentarily motionless target. An alternative open-loop near-minimum time approach is also presented and tested on a 5DOF under-actuated robotic arm. The performance is compared and result shows significant reduction for its time of convergence, to the aforementioned techniques.