Improving visual servoing using fuzzy filters

A new approach to improve visual servoing based on fuzzy filters is proposed in this paper. In visual servoing the speed of convergence depends on a constant gain, that directly influences the velocity of the robot manipulator joints. This velocity can achieve undesired behavior due to the intrinsically discrete visual servoing and when regulator control is used. In this paper, the image features error is filtered by means of fuzzy filters, to improve the behavior of the robot joint velocities. Simulation results are presented for a 6 degrees of freedom (DOF) eye-in-hand system, that demonstrates the efficiency of the proposed approach. Experimental results are also presented when a 2 DOF robot manipulator performs kinematic visual servoing, moving to a pre-defined image features position.