A feedback tracking system for robot (Q1820109)

A causal feedback control law for a specific class of smooth nonlinear multi-input multi-output systems is proposed. Then this law is proved to be inappropriate in order to guarantee tracking capability of the closed- loop system, given a class of reference signals, for which the finite time system convergence condition is satisfied. Next it is proved that the modified feedback control law (no longer a causal-one), together with a local full-rank property of the system output Jacobian, can guarantee the tracking virtue of the closed-loop system, for an arbitrary reference signal in some class. These general remarks are then applied to the control problem specific for manipulators of robots. A manipulator (an open chain of the fifth- order kinematic pairs) is supposed to grasp a target-body moving randomly in \(R^ 6\). The so-called speed-acceleration control algorithm is derived with the property of tracking and grasping an object moving arbitrarily without a priori knowledge of the gripper orientation while grasping. Finally, the results of simulation are given illustrating the case of 9R kinematic structure (with 3R gripper included) of an robot. These results show the convergence of the above tracking algorithm for one particular point in the operating space.