Recursive linearizing and decoupling control of robots

Abstract

A control structure is considered for decoupling and linearizing the dynamic behavior of a robotic manipulator. Since computational efficiency is a crucial consideration in implementation of this control system, a fast recursive algorithm is presented for the necessary digital computations, and the computational requirements are studied in terms of the number of degrees of freedom of a general and open-chain robotic manipulator. An important feature of the algorithm is that decoupling is realized without employing matrix inversion. The sequential recursive algorithm is restructured into a parallel algorithm. A significant improvement in the computational speed is achieved in this manner. The computing requirements of the parallel algorithm are compared with those of the serial algorithm. For a six-degrees-of-freedom robot, the computational cost of the parallel algorithm is approximately 23 % of that of the original serial algorithm. Finally, the processor loads in some regions of the parallel algorithm are redistributed to achieve a balanced scheme. The resulting parallel algorithm requires approximately 17% of the computational effort of the serial algorithm, for a six-degrees-of-freedom robot.