Adaptive regional feedback control of robotic manipulator with uncertain kinematics and depth information

While much progress has been achieved in task-space control of robot, existing task-space sensory feedback control methods fail when the sensor is out of working range. In this paper, we propose an adaptive regional feedback control strategy that enables the robot to start from an initial position outside the field of view and leave the field of view during the movement. The robot kinematics is partitioned into a known internal portion and an unknown external portion. Cartesian-space feedback is used for region reaching control of the known portion and vision feedback is used for tracking control of the unknown portion. The dual feedback information is integrated into a unified controller without designing multiple controllers and switching between them. We shall show that the adaptive controller can transit smoothly from Cartesian-space feedback to vision feedback in the presence of uncertainties in robot dynamics, kinematics and depth information. Experimental results are presented to illustrate the performance of the proposed control method.