Learning and Correcting Robot Trajectory Keypoints from a Single Demonstration

Iñigo Iturrate*, Esben Hallundbæk Østergaard, Martin Rytter, Thiusius Rajeeth Savarimuthu

*Kontaktforfatter for dette arbejde

Publikation: Bidrag til bog/antologi/rapport/konference-proceedingKonferencebidrag i proceedingsForskningpeer review

Resumé

Kinesthetic teaching provides an accessible way for non-experts to quickly and easily program a robot system by demonstration. A crucial aspect of this technique is to obtain an accurate approximation of the robot’s intended trajectory for the task, while filtering out spurious aspects of the demonstration. While several methods to this end have been proposed, they either rely on several demonstrations or on the user explicitly indicating relevant trajectory waypoints. We propose a method, based on the Douglas-Peucker line simplification algorithm that is able to extract the notable
points of a trajectory from a single demonstration. Additionally, by utilizing velocity information in the task space, the method is able to achieve a level of precision that is sufficient for industrial assembly tasks. Along with this, we present a user study that shows that our method enables non-expert robot users to successfully program the robot for an assembly benchmark, and that they find this method intuitive.
OriginalsprogEngelsk
TitelProceedings of the 3rd International Conference on Control, Automation and Robotics
ForlagIEEE
Publikationsdato2017
Sider52-59
ISBN (Trykt)978-1-5090-6089-4
ISBN (Elektronisk)978-1-5090-6088-7, 978-1-5090-6086-3
DOI
StatusUdgivet - 2017
Begivenhed3rd International Conference on Control, Automation and Robotics - Nagoya, Japan
Varighed: 24. apr. 201726. apr. 2017
Konferencens nummer: 3

Konference

Konference3rd International Conference on Control, Automation and Robotics
Nummer3
LandJapan
ByNagoya
Periode24/04/201726/04/2017

Fingeraftryk

Demonstrations
Trajectories
Robots
Teaching

Citer dette

Iturrate, I., Østergaard, E. H., Rytter, M., & Savarimuthu, T. R. (2017). Learning and Correcting Robot Trajectory Keypoints from a Single Demonstration. I Proceedings of the 3rd International Conference on Control, Automation and Robotics (s. 52-59). IEEE. https://doi.org/10.1109/ICCAR.2017.7942660
Iturrate, Iñigo ; Østergaard, Esben Hallundbæk ; Rytter, Martin ; Savarimuthu, Thiusius Rajeeth. / Learning and Correcting Robot Trajectory Keypoints from a Single Demonstration. Proceedings of the 3rd International Conference on Control, Automation and Robotics. IEEE, 2017. s. 52-59
@inproceedings{e1e004a309274433a75f4ce4169347a4,
title = "Learning and Correcting Robot Trajectory Keypoints from a Single Demonstration",
abstract = "Kinesthetic teaching provides an accessible way for non-experts to quickly and easily program a robot system by demonstration. A crucial aspect of this technique is to obtain an accurate approximation of the robot’s intended trajectory for the task, while filtering out spurious aspects of the demonstration. While several methods to this end have been proposed, they either rely on several demonstrations or on the user explicitly indicating relevant trajectory waypoints. We propose a method, based on the Douglas-Peucker line simplification algorithm that is able to extract the notablepoints of a trajectory from a single demonstration. Additionally, by utilizing velocity information in the task space, the method is able to achieve a level of precision that is sufficient for industrial assembly tasks. Along with this, we present a user study that shows that our method enables non-expert robot users to successfully program the robot for an assembly benchmark, and that they find this method intuitive.",
author = "I{\~n}igo Iturrate and {\O}stergaard, {Esben Hallundb{\ae}k} and Martin Rytter and Savarimuthu, {Thiusius Rajeeth}",
year = "2017",
doi = "10.1109/ICCAR.2017.7942660",
language = "English",
isbn = "978-1-5090-6089-4",
pages = "52--59",
booktitle = "Proceedings of the 3rd International Conference on Control, Automation and Robotics",
publisher = "IEEE",
address = "United States",

}

Iturrate, I, Østergaard, EH, Rytter, M & Savarimuthu, TR 2017, Learning and Correcting Robot Trajectory Keypoints from a Single Demonstration. i Proceedings of the 3rd International Conference on Control, Automation and Robotics. IEEE, s. 52-59, 3rd International Conference on Control, Automation and Robotics, Nagoya, Japan, 24/04/2017. https://doi.org/10.1109/ICCAR.2017.7942660

Learning and Correcting Robot Trajectory Keypoints from a Single Demonstration. / Iturrate, Iñigo ; Østergaard, Esben Hallundbæk; Rytter, Martin; Savarimuthu, Thiusius Rajeeth.

Proceedings of the 3rd International Conference on Control, Automation and Robotics. IEEE, 2017. s. 52-59.

Publikation: Bidrag til bog/antologi/rapport/konference-proceedingKonferencebidrag i proceedingsForskningpeer review

TY - GEN

T1 - Learning and Correcting Robot Trajectory Keypoints from a Single Demonstration

AU - Iturrate, Iñigo

AU - Østergaard, Esben Hallundbæk

AU - Rytter, Martin

AU - Savarimuthu, Thiusius Rajeeth

PY - 2017

Y1 - 2017

N2 - Kinesthetic teaching provides an accessible way for non-experts to quickly and easily program a robot system by demonstration. A crucial aspect of this technique is to obtain an accurate approximation of the robot’s intended trajectory for the task, while filtering out spurious aspects of the demonstration. While several methods to this end have been proposed, they either rely on several demonstrations or on the user explicitly indicating relevant trajectory waypoints. We propose a method, based on the Douglas-Peucker line simplification algorithm that is able to extract the notablepoints of a trajectory from a single demonstration. Additionally, by utilizing velocity information in the task space, the method is able to achieve a level of precision that is sufficient for industrial assembly tasks. Along with this, we present a user study that shows that our method enables non-expert robot users to successfully program the robot for an assembly benchmark, and that they find this method intuitive.

AB - Kinesthetic teaching provides an accessible way for non-experts to quickly and easily program a robot system by demonstration. A crucial aspect of this technique is to obtain an accurate approximation of the robot’s intended trajectory for the task, while filtering out spurious aspects of the demonstration. While several methods to this end have been proposed, they either rely on several demonstrations or on the user explicitly indicating relevant trajectory waypoints. We propose a method, based on the Douglas-Peucker line simplification algorithm that is able to extract the notablepoints of a trajectory from a single demonstration. Additionally, by utilizing velocity information in the task space, the method is able to achieve a level of precision that is sufficient for industrial assembly tasks. Along with this, we present a user study that shows that our method enables non-expert robot users to successfully program the robot for an assembly benchmark, and that they find this method intuitive.

U2 - 10.1109/ICCAR.2017.7942660

DO - 10.1109/ICCAR.2017.7942660

M3 - Article in proceedings

SN - 978-1-5090-6089-4

SP - 52

EP - 59

BT - Proceedings of the 3rd International Conference on Control, Automation and Robotics

PB - IEEE

ER -

Iturrate I, Østergaard EH, Rytter M, Savarimuthu TR. Learning and Correcting Robot Trajectory Keypoints from a Single Demonstration. I Proceedings of the 3rd International Conference on Control, Automation and Robotics. IEEE. 2017. s. 52-59 https://doi.org/10.1109/ICCAR.2017.7942660