A robust singularity avoidance method with improved path tracking performance for autonomous manipulation

As an inherent characteristic of articulated manipulator, singularity is an inevitable issue while a manipulator is tracking a path defined in its task-space. In this paper, a new singular task reconstruction (STR) method based on the concept of manipulability ellipsoid is proposed to solve kinematic and algorithmic singularity problems. By projecting the desired task onto the direction of manipulability ellipsoid's semi-axes when the manipulator approaches its singular configuration, the modified task impedes the manipulator to enter its singularity region, and simultaneously guarantees improved path tracking performance as compared to a standard Damper Least-Squares (DLS) method. Moreover, this algorithm can be simply extended in the framework of task-priority based method. Benefited from an online evaluation of singular value and eigenvector of the Jacobian matrix, this new method does not require prior knowledge of an emerging singular configuration. The performance and effectiveness of the proposed STR algorithm is demonstrated by simulation works.