The rolling motion of a soft robotic fingertip is in this paper explicitly included in the control objectives together with the force/position regulation targets. A model based control law is proposed to linearize and decouple the system with respect to the force/position and sliding dynamics based on an appropriately defined task Jacobian. The controller is validated by simulations including rolling on a stationary surface and graspless manipulation of a flat object.

This paper proposes a controller for the stable grasp of an arbitrary-shaped object on the horizontal plane by two robotic fingers with rigid hemispherical fingertips. The controller stabilizes the grasp with optimal force angles and desired finger shaping determined through the choice of a control constant without requiring the utilization of any contact information regarding contact locations and contact angles or any estimates of them. Simulation results demonstrate the performance of the proposed controller and show its clear advantages with respect to other known control schemes.

Dexterity in robot hand object manipulation is irrevocably connected with achieving and maintaining rolling motion and contact at the fingertips. The problem of controlling a robotic fingertip for contact rolling is in this work addressed via the generation of a rolling motion trajectory and the application of a prescribed performance controller that can guarantee contact maintenance and a predefined fast convergence of the rolling tracking error to zero under any contact conditions. A simulation of a five degrees of freedom robot show excellent contact rolling performance even at cases of low friction while alternative controllers lead to contact sliding.

A prescribed performance regulator for the generalized position of the robot arm endpoint in the task space is proposed. The control input which incorporates a transformed error guarantees a prescribed performance regarding the response of the endpoint generalized position error. The use of two different forms of this transformed error will be presented and compared. Mathematical proof of the controller's success in fulfilling the desired goals is given. A simulation of a three degrees of freedom robot is used to confirm the theoretical findings for both cases of the transformed error.