Experimental Evaluation of Synergy-Based In-Hand Manipulation

In this paper, the problem of in-hand dexterous manipulation has been addressed on the base of postural synergies analysis. The computation of the synergies subspace able to represent grasp and manipulation tasks as trajectories connecting suitable configuration sets is based on the observation of the human hand behavior. Five subjects are required to reproduce the most natural grasping configuration belonging to the considered grasping taxonomy and the boundary configurations for those grasps that admit internal manipulation. The measurements on the human hand and the reconstruction of the human grasp configurations are obtained using a vision-based mapping method that assume the kinematics of the robotic hand, used for the experiments, as a simplified model of the human hand. The analysis to determine the most suitable set of synergies able to reproduce the selected grasps and the relative allowed internal manipulation has been carried out. The grasping and in-hand manipulation tasks have been reproduced by means of linear interpolation of the boundary configurations in the selected synergies subspace and the results have been experimentally tested on the UB Hand IV.