Learning of admittance matrix elements for manipulative operations

One approach to the learning of admittance matrix elements in, damping control is presented. It is difficult to determine admittance matrix elements analytically since they depend upon many physical properties, which are not identified easily, of manipulated objects. We propose a method to obtain admittance matrix elements through iterative trials of a manipulative operation. First, mating processes are kinematically analyzed and conditions for admittance matrix elements are derived based on modifiability analysis. Second, reaction forces during a mating process are investigated and a function to evaluate the process is defined. A learning algorithm, is then, developed based on nonlinear programming techniques. The developed algorithm is applied to a mating operation in a computer simulation. Finally, an experimental result is shown to demonstrate the validity of the proposed method