Iterative learning control for whole-arm object manipulation through coordination of torque/velocity-controlled fingers

Abstract

This paper proposes an iterative learning control scheme for whole-arm object manipulation by using a robotic hand system composed of a velocity-controlled finger and a torque-controlled finger. The novel iterative learning control scheme consists of the PD-type iterative learning control scheme for the velocity-controlled finger to follow the desired trajectory and the optimization-based iterative learning control scheme for the torque-controlled finger to maintain equilibrium. The switching between the two controllers is made to achieve compliant manipulation. The validity of the proposed scheme is demonstrated through experiments.