Jamming robotics: Grasping, manipulation, and structure

The development of new materials that are amorphous, adaptive, and programmable holds the promise of tremendous advancements for robotics. We can imagine shape-shifting robots that are soft, reconfigurable, and resilient, addressing many of the problems that robotic platforms face today. In this dissertation we advance the concept of jamming materials (bulk granular materials exploited for their reversible, temperature-independent phase change between a solid-like and liquid-like state), as a possible avenue for reaching this goal. We demonstrate the tremendous potential of jamming materials through applications focusing on robotic end-effectors (hands) used for grasping and manipulation. In addition, we demonstrate the potential for built structures made from jamming materials, and we show how some aspects of jamming material performance can be predicted from measured properties of the grains.;This dissertation is presented in three parts: Grasping, Manipulation, and Structure. In Part I: Jamming-Based Universal Grasping, we detail an approach for achieving a robotic grasping tool that is both extremely simple and extremely capable. By exploiting the jamming phase transition of granular materials, this jamming gripper is a passively conforming device that can rapidly grip and release a wide range of objects. We identify three different gripping modes that a jamming gripper can leverage in order to achieve its gripping function, and we develop and test analytical models for each. We also test a jamming gripper prototype in real-world applications in order to quantify performance on metrics including reliability, error tolerance, holding strength, placement precision, and the ability to “shoot” objects useful distances. Finally, we implement a learning algorithm for improving autonomous grasping capabilities with jamming grippers. The result of this work is one of the simplest and most capable robotic grippers yet developed.;In Part II: Dexterous Robotic Manipulation, we extend our success with jamming grippers to the area of dexterous manipulation (i.e. arbitrary movement of a grasped object within the workspace of the hand). We design a two-fingered hand that incorporates pockets of granular material in the fingertips. This JamHand requires only two actuators and two valves to achieve multiple precision and power grasps, and all six basic dexterous manipulations. We further demonstrate our system on a range of real-world grasping and manipulation challenges, and we successfully modify our prior analytical models to describe a two fingered grasp. The proposed JamHand is the simplest known robotic hand to date that is capable of dexterous manipulation, and, is the first known hand to explicitly demonstrate: precision and power grasping, all six basic dexterous manipulations, and real-world grasping and manipulation challenges. Besides practical application, these results suggest that previously held beliefs may have overestimated the hand complexity required for performing daily tasks.;Part III: Structural Performance of Granular Materials details our testing procedures, equipment, results, and analysis for determining the best suited grains for jamming applications. While the potential applications for jamming materials are now within reach, a working understanding of how to design and build with them remains elusive. Our fundamental understanding of the behavior of jamming materials has not kept pace with the demand for their utilization. By testing a wide variety of potential jamming materials and interrogating the resulting data, our goal is to help propel architects and engineers to greater achievements using jamming materials, during this time when our understanding of their material properties is not yet clear enough to provide much design guidance. Through this process we discover several new trends that can help in differentiating between potential jamming materials based on performance.;This dissertation concludes with an explicit listing of the scientific contributions contained herein, a discussion of the impacts this work has already had on science and engineering education, and an assessment of potential future research directions.