On transferability and contexts when using simulated grasp databases

Research output: Contribution to journalJournal articleResearchpeer-review

Abstract

It has become a common practice to use simulation to generate large databases of good grasps for grasp planning in robotics research. However, the existence of a generic simulation context that enables the generation of high quality grasps that can be used in several different contexts such as bin-picking or picking objects from a table, has to our knowledge not yet been discussed in the literature.

In this paper, we investigate how well the quality of grasps simulated in a commonly used “generic” context transfers to a specific context, both, in simulation and in the real world.

We generate a large database of grasp hypotheses for several objects and grippers, which we then evaluate in different dynamic simulation contexts e.g., free floating (no gravity, no obstacles), standing on a table and lying on a table.

We present a comparison on the intersection of the grasp outcome space across the different contexts and quantitatively show that to generate reliable grasp databases, it is important to use context specific simulation.

We furthermore evaluate how well a state of the art grasp database transfers from two simulated contexts to a real world context of picking an object from a table and discuss how to evaluate transferability into non-deterministic real world contexts.
Original languageEnglish
JournalRobotica
Volume33
Issue number05 (Special Issue)
Pages (from-to)1131-1146
ISSN0263-5747
DOIs
Publication statusPublished - Jun 2015

Cite this

@article{3140b232b1214aa0995591b16ffb86d1,
title = "On transferability and contexts when using simulated grasp databases",
abstract = "It has become a common practice to use simulation to generate large databases of good grasps for grasp planning in robotics research. However, the existence of a generic simulation context that enables the generation of high quality grasps that can be used in several different contexts such as bin-picking or picking objects from a table, has to our knowledge not yet been discussed in the literature.In this paper, we investigate how well the quality of grasps simulated in a commonly used “generic” context transfers to a specific context, both, in simulation and in the real world.We generate a large database of grasp hypotheses for several objects and grippers, which we then evaluate in different dynamic simulation contexts e.g., free floating (no gravity, no obstacles), standing on a table and lying on a table.We present a comparison on the intersection of the grasp outcome space across the different contexts and quantitatively show that to generate reliable grasp databases, it is important to use context specific simulation.We furthermore evaluate how well a state of the art grasp database transfers from two simulated contexts to a real world context of picking an object from a table and discuss how to evaluate transferability into non-deterministic real world contexts.",
author = "J{\o}rgensen, {Jimmy Alison} and Lars-Peter Ellekilde and Dirk Kraft and Petersen, {Henrik Gordon} and Norbert Kr{\"u}ger",
year = "2015",
month = "6",
doi = "10.1017/S0263574714001192",
language = "English",
volume = "33",
pages = "1131--1146",
journal = "Robotica",
issn = "0263-5747",
publisher = "Heinemann",
number = "05 (Special Issue)",

}

On transferability and contexts when using simulated grasp databases. / Jørgensen, Jimmy Alison; Ellekilde, Lars-Peter; Kraft, Dirk; Petersen, Henrik Gordon; Krüger, Norbert.

In: Robotica, Vol. 33, No. 05 (Special Issue), 06.2015, p. 1131-1146.

Research output: Contribution to journalJournal articleResearchpeer-review

TY - JOUR

T1 - On transferability and contexts when using simulated grasp databases

AU - Jørgensen, Jimmy Alison

AU - Ellekilde, Lars-Peter

AU - Kraft, Dirk

AU - Petersen, Henrik Gordon

AU - Krüger, Norbert

PY - 2015/6

Y1 - 2015/6

N2 - It has become a common practice to use simulation to generate large databases of good grasps for grasp planning in robotics research. However, the existence of a generic simulation context that enables the generation of high quality grasps that can be used in several different contexts such as bin-picking or picking objects from a table, has to our knowledge not yet been discussed in the literature.In this paper, we investigate how well the quality of grasps simulated in a commonly used “generic” context transfers to a specific context, both, in simulation and in the real world.We generate a large database of grasp hypotheses for several objects and grippers, which we then evaluate in different dynamic simulation contexts e.g., free floating (no gravity, no obstacles), standing on a table and lying on a table.We present a comparison on the intersection of the grasp outcome space across the different contexts and quantitatively show that to generate reliable grasp databases, it is important to use context specific simulation.We furthermore evaluate how well a state of the art grasp database transfers from two simulated contexts to a real world context of picking an object from a table and discuss how to evaluate transferability into non-deterministic real world contexts.

AB - It has become a common practice to use simulation to generate large databases of good grasps for grasp planning in robotics research. However, the existence of a generic simulation context that enables the generation of high quality grasps that can be used in several different contexts such as bin-picking or picking objects from a table, has to our knowledge not yet been discussed in the literature.In this paper, we investigate how well the quality of grasps simulated in a commonly used “generic” context transfers to a specific context, both, in simulation and in the real world.We generate a large database of grasp hypotheses for several objects and grippers, which we then evaluate in different dynamic simulation contexts e.g., free floating (no gravity, no obstacles), standing on a table and lying on a table.We present a comparison on the intersection of the grasp outcome space across the different contexts and quantitatively show that to generate reliable grasp databases, it is important to use context specific simulation.We furthermore evaluate how well a state of the art grasp database transfers from two simulated contexts to a real world context of picking an object from a table and discuss how to evaluate transferability into non-deterministic real world contexts.

U2 - 10.1017/S0263574714001192

DO - 10.1017/S0263574714001192

M3 - Journal article

VL - 33

SP - 1131

EP - 1146

JO - Robotica

JF - Robotica

SN - 0263-5747

IS - 05 (Special Issue)

ER -