Automation of grasp planning for a dextrous hand

An approach is presented for the automatic preshaping of a dextrous multifingered hand. Due to the high number of degrees of freedom, it is impossible to build and to explore the whole set of possible configurations of the gripper in order to choose a single way of grasping compatible with the object shape, as it is classically done for simple parallel jaws grippers. Research in neurophysiology has shown that during the reaching of an object, the human hand preshapes in order to prepare the matching between the hand and object shapes; a preshape represents a class of configurations, among which the exact grasping configuration is chosen from tactile and force feedback coming from the fingers at grasping time. The problem is studied for three types of preshape, the encompassing preshape, the precision preshape and the lateral preshape.<
>