Precise In-Hand Motion Control of Objects Using Soft Actuators and Visual Feedback

Abstract

Soft robots are expected to resolve problems faced by highly rigid robots, grasping objects of varied shape, for example. In this paper, we propose a soft robotic hand that grasps objects of various shapes and realizes high-precision manipulation of the grasped objects. This soft robotic hand has four pneumatically-driven soft actuators in its grasp and these soft actuators are made on a multi-material 3D printer which can print in both soft and hard materials. For high-precision object manipulation, we also propose a robust visual feedback control. This control is realized by estimating a linear relationship between forces/torques generated on the grasped object and inputs of the soft actuators. By using this proposed method, vertical plane manipulation of the grasped object is realized. As a result, position and orientation errors, when using this method, are about 0.30 mm and about 0.03 rad.