Extrinsic dexterity: In-hand manipulation with external forces

By Nikhil Chavan Dafle, Alberto Rodriguez, Robert Paolini, Bowei Tang, Siddhartha S. Srinivasa, M. Erdmann, Matthew Thomas Mason, Ivan Lundberg, H. Staab, T. Fuhlbrigge. 2014

313 citations recorded by the source corpus.

“In-hand manipulation” is the ability to reposition an object in the hand, for example when adjusting the grasp of a hammer before hammering a nail. The common approach to in-hand manipulation with robotic hands, known as dexterous manipulation [1], is to hold an object within the fingertips of the hand and wiggle the fingers, or walk them along the object's surface. Dexterous manipulation, however, is just one of the many techniques available to the robot. The robot can also roll the object in the hand by using gravity, or adjust the object's pose by pressing it against a surface, or if fast enough, it can even toss the object in the air and catch it in a different pose. All these techniques have one thing in common: they rely on resources extrinsic to the hand, either gravity, external contacts or dynamic arm motions. We refer to them as “extrinsic dexterity”. In this paper we study extrinsic dexterity in the context of regrasp operations, for example when switching from a power to a precision grasp, and we demonstrate that even simple grippers are capable of ample in-hand manipulation. We develop twelve regrasp actions, all open-loop and hand-scripted, and evaluate their effectiveness with over 1200 trials of regrasps and sequences of regrasps, for three different objects (see video [2]). The long-term goal of this work is to develop a general repertoire of these behaviors, and to understand how such a repertoire might eventually constitute a general-purpose in-hand manipulation capability.

Subfields: Biomedical Engineering, Control and Systems Engineering

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biomedical Engineering

In Control and Systems Engineering

Citations in the corpus

Solving Rubik's Cube with a Robot Hand (2019, background): “This includes rolling [10, 41, 42, 15, 26], sliding [15, 100], finger gaiting [42], finger tracking [90], pushing [24], and re-grasping [109, 25].”
Self-Supervised Correspondence in Visuomotor Policy Learning (2019, background): “The “Push-then-grab plate” task as well is highly challenging given the visual clutter, the symmetry and lack of visual texture for the object, and requires using “extrinsic dexterity” [39] via the wood block to enable sliding the gripper into position to grasp the plate.”
Learning dexterous in-hand manipulation (2018, background): “This includes rolling [8, 22, 23, 9, 13], sliding [9, 59], ﬁnger gaiting [23], ﬁnger tracking [51], pushing [11], and re-grasping [65, 12].”
TossingBot: Learning to Throw Arbitrary Objects With Residual Physics (2019, background): “deos are available athttp://tossingbot.cs.princeton.edu I. INTRODUCTION Throwing is a means to increase the capabilities of a manipulator by exploiting dynamics, a form of dynamic extrinsic dexterity [5]. In the case of pick-and-place, throwing enables a robot arm to place objects rapidly into boxes”
Differentiable Physics and Stable Modes for Tool-Use and Manipulation Planning (2018, background): “The classical manipulation literature emphasizes the importance of stable interaction modes and funnels [19, 17, 3].”
Rubik's Cube Handling Using a High-Speed Multi-Fingered Hand and a High-Speed Vision System (2018, background): “realized 12 types of regrasping by using grippers with low degrees of freedom and external forces such as gravity and inertia [3].”
Dynamic in-hand sliding manipulation (2015, background): “A manually designed example of this can be found in [3].”
Analysis and Observations From the First Amazon Picking Challenge (2016, background): “…their poor ability to manipulate an object; this capability was not prominent in the APC so far, however future contests are expected to have more populated bins that may require more ﬂexible manipulation and object rearrangement [37], or a more explicit exploitation of the environment [38], [39].”
A System for General In-Hand Object Re-Orientation (2021, background): “In contrast to paralleljaw grasping, pushing, pivoting [32], or pick-and-place, dexterous manipulation typically involves continuously controlling force to the object through the fingertips of a robotic hand [33].”
Learning Visual Affordances for Robotic Manipulation (2019, background): “Throwing is a means to increase the capabilities of a manipulator by exploiting dynamics, a form of dynamic extrinsic dexterity [15].”
Stable Prehensile Pushing: In-Hand Manipulation with Alternating Sticking Contacts (2017, background): “In our previous work, we presented extrinsic dexterity – an approach to in-hand manipulation using gravity, dynamic motions of the arm and contacts with the environment [8].”
Learning Dexterous In-Hand Manipulation. (2018, background): “This includes rolling [8, 22, 23, 9, 13], sliding [9, 59], finger gaiting [23], finger tracking [51], pushing [11], and re-grasping [65, 12].”
Sampling-based Planning of In-Hand Manipulation with External Pushes (2017, background): “Diverging from this assumption, our recent work demonstrated the use of gravity, dynamic motions, and contacts with the environment to regrasp objects using a library of hand-scripted motions [3].”
Feedback Control of the Pusher-Slider System: A Story of Hybrid and Underactuated Contact Dynamics (2016, methodology): “[8, 9] demonstrated that simple robotic hands can be used to perform fast and effective regrasp strategies by using the robotic a rm and the environment as an external source of actuation.”
Reinforcement Learning for Pivoting Task (2017, background): “ gripper. Pivoting consists of rotating an object or a tool between two ﬁngers to reorient it to a desired angle. This problem falls in the general class of extrinsic dexterity problems formulated by [1]. Since many tasks in robotics require interaction with objects and tool use, the ability of plac”
Realtime State Estimation with Tactile and Visual Sensing. Application to Planar Manipulation (2017, methodology): “In the future, we would like to (1) test our estimation in the context of a pushing controller [23] for reactive manipulation, which has been tested with accurate ground truth feedback from an external tracking system, and (2) generalize to 3D tasks such as prehensile manipulation [24, 25, 26].”
Trends and challenges in robot manipulation (2019, background): “Extrinsic dexterity is the ability to compensate for the lack of degrees of freedom by using external support, such as friction, gravity, and contact surfaces (69).”