A Technique Based on Adaptive Extended Jacobians for Improving the Robustness of the Inverse Numerical Kinematics of Redundant Robots

The extended Jacobian is a technique for solving the redundancy of redundant robots. It is based on the definition of secondary tasks, through constraint functions that are added to the mapping between joint rates and end-effector's twist. Several approaches showed its potential, applications, and limitations. In general, the constraint functions are a linear combination of basic functions with constant coefficients. This paper proposes the use of adaptive coefficients in such functions by using the conditioning index of the extended Jacobian as a quality measure. A good conditioning index of the extended Jacobian keeps the robot far from singularities and contributes to the solution of the inverse kinematics. In this paper, initially, the extended Jacobian and the proposed algorithm are discussed, and then, two tests in different circumstances are presented in order to validate the proposal.