抄録
A new control method named “Redundancy Decomposition Control”, is proposed for the resolved acceleration control of redundant manipulator. The proposed method can be summarized as follows: (1) The redundant degree of freedom of the given manipulator is decomposed into a non-redundant subset. (2) For every time interval from t to t+Δt, the acceleration of the corresponding subset of degree of freedom is assumed to be controlled and the inverse dynamics as well as evaluation function are calculated using the non-redundant Jacobian inverse matrix. The calculation is parallelly performed for all the combinations of degree of freedom. (3) The optimum set of degree of freedom is selected and it is used for the control of the time interval from t to t+Δt.
Comparison is made between the proposed method, and conventional methods, that is the pseudo-inverse matrix method and the null-space vector method. It is shown that the proposed method almost always realizes better motion planning than the former two methods, and has the capability to be used in real time control.
