Learning a State Transition Model of an Underactuated Adaptive Hand

Avishai Sintov, Andrew S. Morgan, Andrew Kimmel, Aaron M. Dollar, Kostas Bekris, Abdeslam Boularias

Research output: Contribution to journalArticle

Abstract

Fully actuated multifingered robotic hands are often expensive and fragile. Low-cost underactuated hands are appealing but present challenges due to the lack of analytical models. This letter aims to learn a stochastic version of such models automatically from data with minimum user effort. The focus is on identifying the dominant, sensible features required to express hand state transitions given quasi-static motions, thereby enabling the learning of a probabilistic transition model from recorded trajectories. Experiments both with Gaussian processes (GP) and neural etwork models are included for analysis and evaluation. The metric for local GP regression is obtained with a manifold learning approach, known as Diffusion Maps, to uncover the lower-dimensional subspace in which the data lies and provide a geodesic metric. Results show that using Diffusion Maps with a feature space composed of the object position, actuator angles, and actuator loads, sufficiently expresses the hand-object system configuration and can provide accurate enough predictions for a relatively long horizon. To the best of the authors' knowledge, this is the first learned transition model for such underactuated hands that achieves this level of predictability. Notably, the same feature space implicitly embeds the size of the manipulated object and can generalize to new objects of varying sizes. Furthermore, the learned model can identify states that are on the verge of failure and which should be avoided during manipulation. The usefulness of the model is also demonstrated by integrating it with closed-loop control to successfully and safely complete manipulation tasks.

Original languageEnglish (US)
Article number8624443
Pages (from-to)1287-1294
Number of pages8
JournalIEEE Robotics and Automation Letters
Volume4
Issue number2
DOIs
StatePublished - Apr 1 2019

Fingerprint

Transition Model
State Transition
End effectors
Feature Space
Gaussian Process
Manipulation
Actuator
Express
Metric
Manifold Learning
Closed-loop Control
Predictability
Actuators
Probabilistic Model
Model
Analytical Model
Geodesic
Robotics
Horizon
Regression

All Science Journal Classification (ASJC) codes

  • Control and Systems Engineering
  • Human-Computer Interaction
  • Biomedical Engineering
  • Mechanical Engineering
  • Control and Optimization
  • Artificial Intelligence
  • Computer Science Applications
  • Computer Vision and Pattern Recognition

Keywords

  • Dexterous Manipulation
  • Tendon/Wire Mechanism
  • Underactuated Robots

Cite this

Sintov, Avishai ; Morgan, Andrew S. ; Kimmel, Andrew ; Dollar, Aaron M. ; Bekris, Kostas ; Boularias, Abdeslam. / Learning a State Transition Model of an Underactuated Adaptive Hand. In: IEEE Robotics and Automation Letters. 2019 ; Vol. 4, No. 2. pp. 1287-1294.
@article{54808fb992494a758961b2b452e4cbdb,
title = "Learning a State Transition Model of an Underactuated Adaptive Hand",
abstract = "Fully actuated multifingered robotic hands are often expensive and fragile. Low-cost underactuated hands are appealing but present challenges due to the lack of analytical models. This letter aims to learn a stochastic version of such models automatically from data with minimum user effort. The focus is on identifying the dominant, sensible features required to express hand state transitions given quasi-static motions, thereby enabling the learning of a probabilistic transition model from recorded trajectories. Experiments both with Gaussian processes (GP) and neural etwork models are included for analysis and evaluation. The metric for local GP regression is obtained with a manifold learning approach, known as Diffusion Maps, to uncover the lower-dimensional subspace in which the data lies and provide a geodesic metric. Results show that using Diffusion Maps with a feature space composed of the object position, actuator angles, and actuator loads, sufficiently expresses the hand-object system configuration and can provide accurate enough predictions for a relatively long horizon. To the best of the authors' knowledge, this is the first learned transition model for such underactuated hands that achieves this level of predictability. Notably, the same feature space implicitly embeds the size of the manipulated object and can generalize to new objects of varying sizes. Furthermore, the learned model can identify states that are on the verge of failure and which should be avoided during manipulation. The usefulness of the model is also demonstrated by integrating it with closed-loop control to successfully and safely complete manipulation tasks.",
keywords = "Dexterous Manipulation, Tendon/Wire Mechanism, Underactuated Robots",
author = "Avishai Sintov and Morgan, {Andrew S.} and Andrew Kimmel and Dollar, {Aaron M.} and Kostas Bekris and Abdeslam Boularias",
year = "2019",
month = "4",
day = "1",
doi = "10.1109/LRA.2019.2894875",
language = "English (US)",
volume = "4",
pages = "1287--1294",
journal = "IEEE Robotics and Automation Letters",
issn = "2377-3766",
number = "2",

}

Learning a State Transition Model of an Underactuated Adaptive Hand. / Sintov, Avishai; Morgan, Andrew S.; Kimmel, Andrew; Dollar, Aaron M.; Bekris, Kostas; Boularias, Abdeslam.

In: IEEE Robotics and Automation Letters, Vol. 4, No. 2, 8624443, 01.04.2019, p. 1287-1294.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Learning a State Transition Model of an Underactuated Adaptive Hand

AU - Sintov, Avishai

AU - Morgan, Andrew S.

AU - Kimmel, Andrew

AU - Dollar, Aaron M.

AU - Bekris, Kostas

AU - Boularias, Abdeslam

PY - 2019/4/1

Y1 - 2019/4/1

N2 - Fully actuated multifingered robotic hands are often expensive and fragile. Low-cost underactuated hands are appealing but present challenges due to the lack of analytical models. This letter aims to learn a stochastic version of such models automatically from data with minimum user effort. The focus is on identifying the dominant, sensible features required to express hand state transitions given quasi-static motions, thereby enabling the learning of a probabilistic transition model from recorded trajectories. Experiments both with Gaussian processes (GP) and neural etwork models are included for analysis and evaluation. The metric for local GP regression is obtained with a manifold learning approach, known as Diffusion Maps, to uncover the lower-dimensional subspace in which the data lies and provide a geodesic metric. Results show that using Diffusion Maps with a feature space composed of the object position, actuator angles, and actuator loads, sufficiently expresses the hand-object system configuration and can provide accurate enough predictions for a relatively long horizon. To the best of the authors' knowledge, this is the first learned transition model for such underactuated hands that achieves this level of predictability. Notably, the same feature space implicitly embeds the size of the manipulated object and can generalize to new objects of varying sizes. Furthermore, the learned model can identify states that are on the verge of failure and which should be avoided during manipulation. The usefulness of the model is also demonstrated by integrating it with closed-loop control to successfully and safely complete manipulation tasks.

AB - Fully actuated multifingered robotic hands are often expensive and fragile. Low-cost underactuated hands are appealing but present challenges due to the lack of analytical models. This letter aims to learn a stochastic version of such models automatically from data with minimum user effort. The focus is on identifying the dominant, sensible features required to express hand state transitions given quasi-static motions, thereby enabling the learning of a probabilistic transition model from recorded trajectories. Experiments both with Gaussian processes (GP) and neural etwork models are included for analysis and evaluation. The metric for local GP regression is obtained with a manifold learning approach, known as Diffusion Maps, to uncover the lower-dimensional subspace in which the data lies and provide a geodesic metric. Results show that using Diffusion Maps with a feature space composed of the object position, actuator angles, and actuator loads, sufficiently expresses the hand-object system configuration and can provide accurate enough predictions for a relatively long horizon. To the best of the authors' knowledge, this is the first learned transition model for such underactuated hands that achieves this level of predictability. Notably, the same feature space implicitly embeds the size of the manipulated object and can generalize to new objects of varying sizes. Furthermore, the learned model can identify states that are on the verge of failure and which should be avoided during manipulation. The usefulness of the model is also demonstrated by integrating it with closed-loop control to successfully and safely complete manipulation tasks.

KW - Dexterous Manipulation

KW - Tendon/Wire Mechanism

KW - Underactuated Robots

UR - http://www.scopus.com/inward/record.url?scp=85065929850&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85065929850&partnerID=8YFLogxK

U2 - 10.1109/LRA.2019.2894875

DO - 10.1109/LRA.2019.2894875

M3 - Article

VL - 4

SP - 1287

EP - 1294

JO - IEEE Robotics and Automation Letters

JF - IEEE Robotics and Automation Letters

SN - 2377-3766

IS - 2

M1 - 8624443

ER -