Impedance Control of Robot Manipulator in Contact Task Using Machine Learning

In performing contact tasks using robot manipulators, force control is essential. One approach is to select appropriate stiffness ellipse at the endpoint of the manipulator, where stiffness ellipse is a geometrical shape of force element represented in the principal axis of task space. In this study, we introduce a novel method to tailor stiffness ellipse required to perform contact tasks by using associative search network. Using appropriate performance indexes in the open-door task experiment, we acquired stiffness ellipse trajectory which optimizes dynamic movement of manipulator. Derived stiffness ellipse (or impedance in general) through learning process can be used for the similar task of learning process