Decentralized cooperation control of redundant manipulators

This paper proposes three decentralized cooperation controllers for multiple redundant manipulators handling the same object. The controllers control position of the object by a position feedback in the operational space through pseudo-inverses of the Jacobians, but they control the internal forces in different ways. The first controller uses only a feedforward of the desired force, but the others include a force feedback. The first two require accurate models of the robots and guarantee globally and exponentially asymptotic convergence of both the position and force errors. Although only the asymptotic convergence is assured in the third controller, the robotic models are not needed. The controllers require neither a coordinator nor communication of the joint angles and velocities subject to that the robots share the knowledge of the desired trajectory of the object and the desired force.