Abstract
While the modelling analysis of the kinetostatic behaviour of underactuated tendon-driven robotic fingers has been largely addressed in the literature, tendon routing is often not considered by these theoretical models. Tendon routing path plays a fundamental role in defining joint torques, and subsequently the force vectors produced by the phalanxes. However, dynamic tendon behaviour is difficult to predict and is influenced by many external factors including tendon friction, the shape of the grasped object, the initial pose of the fingers, and finger contact points. In this paper, we present an experimental comparison of the force performance of nine fingers, with different tendon routing configurations. We use the concept of force-isotropy, in which forces are equal and distributed on each phalanx as the optimum condition for an adaptive grasp. Our results show only some of the finger designs surveyed exhibited a partial adaptive behaviour, showing distributed force for the proximal and distal phalanxes throughout grasping cycles, while other routings resulted in only a single finger remaining in contact with the object.