(An) adaptive hybrid position/force control of manipulators in task-oriented coordinates

Abstract

Hybrid position/force control is a form of manipulator control in which the degrees of freedom of manipulators are partitioned into a position controlled subset and a force controlled subset. This hybrid position/force control is an effective tool for carrying out the task whose geometry constrains position of the manipulator. This study organizes a task-oriented architecture of the hybrid control under the consideration of dynamic characteristics of the manipulator and robustness to its environmental uncertainties: A type of SISO pole assignment self-tuners was adopted as position and force controllers based upon results of dynamic coupling evaluation that the cartesian space dynamics is less coupled compared with the joint space dynamics. Controller formulation in task-oriented coordinates has an additional benifit that there is no need to solve the inverse kinematic problems. An output synthesizing method was developed also in the task-oriented frame, which utilizes only joint encoders and wrist force sensor. To demonstrate the validities of the architecture, various example tasks were numerically experimented using the manipulator which has the kinematic and dynamic characteristics analogous to ``PUMA 560``. Although they were the somewhat idealized tasks, stimulative results could be obtained in the various task specifications.