Interpolation control posture design for in-hand manipulation

Abstract

Object handling and manipulation are important functions of dexterous anthropomorphic hands. As it is difficult to create a manipulation model analytically for performing complex tasks as humans do, a simple way to accomplish generic in-hand manipulation is desirable for robotic hands, especially for robot hands with few sensors. In this paper, we designed standard grasping postures based on the grasping behavior of humans. Interpolation control is used to control the robot hand. The combination of these postures results in versatile object manipulation. This control strategy is evaluated by testing it with the 16 degrees of freedom Allegro Hand grasping different sizes of spheres.