Soft Materials Architectures for Robot Manipulation

Abstract

Robot manipulation has been a prolific subject of academic research for several decades - however, today's robotic manipulators have yet to demonstrate an ability to perform robust and versatile dexterous manipulation. This challenge can largely be attributed to a trade-off between complexity and capability in the design of robotic manipulators. In order to bridge this gap, we leverage the paradigm of Soft Robotics. This work involves the development of two distinct soft robotic end effectors. The first is a stiffness and adhesion tuning gripper, which is capable of grasping objects of various shapes, weights, and sizes. Pull-off tests conducted using this gripper show that transitioning from a soft to stiff state during grasping enables up to 6x increase in adhesion strength. The second end effector developed for this study is a tendon driven foam hand, which is intended to be highly dexterous while requiring a low number of motors for actuation. We develop a hand which demonstrates a high potential for dexterity, and indicate a design based on postural synergies which would allow it to complete 5 distinct dexterous tasks using only two motors.