MOTION RECOVERY OF PARALLEL MANIPULATORS USING GENERALIZED INVERSE AND PARTITIONED JACOBIAN MATRIX

In this paper, the motion recovery of parallel manipulators due to joint failure is addressed. The failed joints have either zero velocities, locked joints, or nonzero velocities which are different from the desired velocities. For the Jacobian matrix of the failed leg, its generalized inverse, which satisfies the first, second, and fourth equations among four equations of Penrose, is utilized to characterize the minimum norm of the correctional velocity vector related to the healthy joints. The platform twist is fully recovered if the Jacobian matrix of the failed leg is of full row- rank or if the platform twist is in the range space of the Jacobian matrix of the failed leg. When the platform twist is in the orthogonal complement of the range space of the Jacobian matrix of the failed leg, or when the Jacobian matrix of the failed leg is not of full row-rank, the partial recovery of the platform twist is achieved using the partitioned Jacobian matrix, while the correctional joint velocity after the failure is minimized.