Comparison of object-level grasp controllers for dynamic dexterous manipulation

Object-level control of a dexterous robot hand provides an intuitive high-level interface to solve fine manipulation tasks. In the past, many algorithms were proposed based on a weighted pseudoinverse of the grasp map. In a different approach, Stramigioli introduced a virtual-object based controller-called an Intrinsically Passive Controller (IPC). These controllers are reviewed and compared. A new controller that is similar to the IPC but using a virtual frame rather than a virtual object is subsequently proposed. The controllers are compared with respect to their object force distribution, the extendability to N fingers, the ease of specifying the object-level impedance and grasp forces, the dimensionality of the coupling springs, the internal controller dynamics, and the computational effort. Controllers for robotic hands usually implement only stiffness controllers and do not program the damping. We address how to choose and implement damping as a function of the desired object-level stiffness and the effective hand-object inertia. The evaluation reveals the dynamic effects of fast motions, which should not be neglected for the design of grasp controllers in practice. The application of the controllers to the torque-controlled DLR Hand II is employed to compare their effectiveness in experiments.