抄録
In recent years, the development of industry diversifies tasks for an industrial robot. To meet with various tasks, industrial robots get complicated. The complexity causes a decrease in productivity due to its poor capability of responding to a changing in task contents and an increase in calculation amount in control. As a method to solve these problems, there is the method called modularization, that treats a system by dividing into modules. The method simplifies reconstruction of systems for changing tasks and improves calculation speed by allocating calculations for each module. However, when a modularized system is reconstructed, the structure of the system changes. The changing the structure requires recalculation of the transposed Jacobian matrix utilized for force control in accordance with change in the structure of the system. The recalculation cannot fully take the advantage possessed by modularization of easiness in reconstructing systems. In this research, we propose a method of controlling force while maintaining the advantage of easiness in reconstructing a system by utilizing modularized inverse dynamics model instead of transposed Jacobian matrix. The method is inspired with the fact that an inverse dynamics model can be treated as a substitute for the transposed Jacobian matrix. In addition, this research shows that the proposed method can maintain performance comparable to that of the conventional method by conducting an actual machine experiment utilizing a multi-link robot.
