Abstract
The solution set of nonlinear kinematic constraint equations can be divided into regular and critical solution subsets, according to linear dependency of gradient vectors of the constraint equations. However, relative to a given input coordinate set, the solution set can also be separated into singular and nonsingular solution subsets, according to whether the sub-Jacobian matrix with respect to the output and intermediate coordinates is rank deficient or not. By providing precise definitions and classifications of singular configurations, from both mathematical and physical points of view, a better understanding of the kinematic behavior of singularity is obtained. Moreover, by extending the definition of the singular solution set to the output space and exploring the mathematical meaning of it, the difficulty in formulating mathematical expressions for workspace problems is resolved.