Source-linked AI summary
A Compliant, Underactuated Hand for Robust Manipulation
Lael U. Odhner, Leif P. Jentoft, Mark R. Claffee, Nicholas Corson, Yaroslav Tenzer, Raymond R. Ma, Martin Buehler, Robert Kohout, Robert D. Howe, Aaron M. Dollar
TL;DR
Practical robot hands need to be compact, inexpensive, robust, and capable of more than one or two grasping configurations. This paper introduces a five-actuator i-HY Hand designed bottom-up around representative tasks, using underactuation and passive compliance for grasping and manipulation. Experiments demonstrate a broad range of grasping and manipulation capabilities, while measured fingertip stiffness supports stable grasping and passive force adjustment.
Problem
Existing robot hands often trade practical cost, robustness, and simplicity against broad grasping and manipulation capabilities.
Method
The i-HY Hand is designed bottom-up from representative tasks and manipulation primitives, using five actuators, underactuation, and passive compliant fingers.
Results
The hand performs power and fingertip grasps, simple in-hand manipulation, stable fingertip grasping, and passive adjustment of grasp forces.
Takeaways & Limitations
Task-centric design can produce a medium-complexity hand that is robust, easy to control, inexpensive, and capable of a wide range of tasks.
Takeaways & Limitations
The demonstrated task scope includes specific representative requirements, such as accessing locks and operating a radio push-to-talk button, rather than unrestricted general-purpose manipulation.
Abstract
from arXiv · showhide
This paper introduces the i-HY Hand, an underactuated hand driven by 5 actuators that is capable of performing a wide range of grasping and in-hand manipulation tasks. This hand was designed to address the need for a durable, inexpensive, moderately dexterous hand suitable for use on mobile robots. The primary focus of this paper will be on the novel minimalistic design of i-HY, which was developed by choosing a set of target tasks around which the design of the hand was optimized. Particular emphasis is placed on the development of underactuated fingers that are capable of both firm power grasps and low- stiffness fingertip grasps using only the passive mechanics of the finger mechanism. Experimental results demonstrate successful grasping of a wide range of target objects, the stability of fingertip grasping, as well as the ability to adjust the force exerted on grasped objects using the passive finger mechanics.
1 Introduction
The paper addresses the gap between complex research hands and practical robot hands by introducing i-HY, a five-actuator hand designed bottom-up around useful grasping and manipulation tasks.
- Practical robot hands must be inexpensive, compact, robust, and capable of grasping, in-hand transitions, and basic tool use.
- Top-down design principles can produce hands with many actuators, elaborate sensing, high cost, fragility, and difficult control.
- i-HY uses a bottom-up process that begins by analyzing required tasks and culminates in a minimalistic hand design.
- The i-HY Hand has five actuators and performs power grasping, fingertip grasping, and simple in-hand manipulation.
2 Task Analysis and Hand Design
i-HY was designed by reducing representative manipulation tasks to grasping and manipulation primitives, then implementing them with a three-finger, five-actuator hand using passive compliance, underactuation, and modular sensing.
- Task Analysis and Hand Design: Bottom-up hand design specifies the hand from required tasks rather than high-level mathematical sufficient conditions.The approach distills tasks into primitives and uses them to determine layout, actuation, and finger structure.
- Task Requirements: The task set includes access, tool use, object handling, and grasping unknown objects, including keys, zippers, wire cutters, markers, drills, radios, flashlights, hammers, and heavy objects.
- Task Requirements: The required grasp repertoire includes cylindrical, spherical, opposed pinch, cylindrical power, spherical power, and lateral grasps.The lateral grasp rotates two fingers normal to their motion plane so a third finger can push against them.
- Task Requirements: Key manipulation primitives include pinch-to-power transfer, rotating thin objects into power grasps, pivoting objects with a third finger, and operating triggers or buttons.
- Hand Structure and Actuation: The hand uses three fingers in a triangular layout, with two fingers coupled for adduction and abduction to switch among opposed, spherical, and interlaced configurations.
- Hand Structure and Actuation: Five actuators provide individual finger flexion, thumb planar positioning, and coupled motion of the two underactuated fingers.
- Finger Design: Compliant flexure joints provide three-dimensional fingertip motion, passive adaptation, and extension when flexor tendons are relaxed.The fingers use a distal elastomer flexure and a proximal torsion spring, while sealed construction improves resistance to water, dirt, and impact.
- Sensing: The sensing system combines magnetic encoders, optical flexure sensors, accelerometers, and MEMS tactile arrays for kinematics and contact detection.
3 Compliant Underactuated Fingers
The i-HY fingers use passive compliance and underactuation to support both robust power grasps and stable, gentle fingertip grasps. Their geometry and mechanics produce predictable force responses without requiring specialized low-impedance actuators or additional control mechanisms.
- 3.1 Designing Compliant Fingers for Robust Power Grasps: A power grasp proceeds through sweeping, followed by caging as distal links flex around the object.The underactuated finger model represents proximal and distal joint motion through a constrained tendon and elastic joint mechanics.
- 3.1 Designing Compliant Fingers for Robust Power Grasps: After distal contact, added constraints stiffen the hand and produce a larger increase in object force for further tendon excursion.During caging, flexure bending builds force slowly; distal contact changes the force-excursion slope, after which force rises approximately linearly.
- 3.2 Extending Compliance to Stable Fingertip Grasps: Compliance at a translated point is obtained from the rigid-body compliance transformation, whose blocks represent Cartesian, torsional, and coupled compliance.The analysis maps compliance along the distal link to identify fingertip locations and directions with the desired behavior.
- 3.2 Extending Compliance to Stable Fingertip Grasps: The finger’s principal compliance is designed to align approximately normal to the fingertip, enabling gentle motion while retaining lateral grasp stability.The distal link was shortened so contact occurs away from its center of compliance, where fingertip compliance is significant and principally normal to the contact surface.
- 3.2 Extending Compliance to Stable Fingertip Grasps: Fingertip force rises gradually and linearly with tendon excursion until distal-link travel is constrained, after which the relationship stiffens predictably.This behavior makes the finger act like a passive series elastic actuator for fingertip grasping.
- 3.3 Summary: The resulting fingers preserve adaptive power grasping while adding low-stiffness fingertip grasping through passive mechanics and predictable force-excursion behavior.No ad hoc fingertip mechanism is needed because compliance is tuned through the finger-link geometry and passive underactuated mechanics.
4 Grasping Primitives in Task Space
The i-HY hand uses compliance and underactuation to support robust grasping, passive adaptation, adjustable fingertip stiffness, and in-hand manipulation across practical tasks.
- Compliance: Compliance protects the fingers during unintended table contact while allowing adaptation to object geometry during bottle-lid tightening.The fingers deform without damage during collisions and passively adjust during tightening.
- Power grasp acquisition: Autonomous power-grasp acquisition used Kinect-estimated object centroids and principal axes, followed by palm positioning and a prerecorded closing trajectory.The same strategy enabled grasping across a series of objects placed on a table.
- Power grasp acquisition: More than 19 of 20 attempts succeeded for one tested object, while the small pen was grasped successfully in 12 of 20 attempts because point-cloud errors reduced positioning accuracy.The pen’s small size caused larger Kinect point-cloud errors and less accurate pre-positioning.
- Adaptive fingertip mechanics: Distal flexures provide passive fingertip adaptation, while flexing the distal link toward its travel limit increases torsional stiffness for tasks such as key turning.This mechanical adjustment allows larger moments to be applied to the key.
- Grasp stability: Spherical pinch grasps remained stable during marker writing and forceful battery collisions with a table, despite substantial finger deformation.The examples show deformation accompanying external contact without object loss.
- Grasp stability: Measured fingertip-grasp stiffness was well-conditioned, with no atypically stiff or compliant directions, while hysteresis reflected tendon friction and viscoelastic materials.An apple weighing approximately 100 g would produce about 2 mm of deflection at 0.5 N/mm stiffness.
- In-hand manipulation: Compliance also supports in-hand manipulation, including three-dimensional transfer of an AA battery from a table into a flashlight while maintaining grasp forces.The hand can transition between opposed and spherical pinch grasps while retaining the object.
5 Conclusions
The i-HY Hand combines five actuators with a simple, compliant, underactuated design to perform diverse grasping and manipulation tasks. Its task-centric design supports robust object handling, tool use, and transitions into stable grasps.
- Conclusions: The five-actuator i-HY Hand performs diverse tasks while remaining simple, robust, easy to control, and inexpensive.Its demonstrated capabilities include lifting tools, operating trigger-equipped tools, and moving keys into stable pinch grasps.