An iterative robot-image Jacobian approximation of image-based visual servoing for joint limit avoidance

In this paper, we propose a new method addressing the robot-image Jacobianapproximation of image-based visual servoing (IBVS) for a redundant manipulator. Therobot-image Jacobian is approximated iteratively and is a model free. A linearised model of therobot-image Jacobian is applied, based on the first order Taylor series approximation. A weightedleast norm solution is induced in a pseudo inverse computation of the approximated robot-imageJacobian. The resulting control law then can be used for visual servoing tasks with joint limitavoidance capability using both a static target or moving target. The self-motion of the robotjoints resolves the redundancy during visual servoing tasks when one or more joints areapproaching their joint limits.A design and stability analysis of the proposed method is discussed in this paper.Simulated and real-time experiments using a 7 DOF PowerCube robot manipulator areconducted. The IBVS is configured using a monovision eye-in-hand system configuration. Thesystem behaviour and performances of the proposed method are presented and analysed.