Learning dexterous grasps that generalise to novel objects by combining hand and contact models

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Standard

Learning dexterous grasps that generalise to novel objects by combining hand and contact models. / Kopicki, Marek; Detry, Renaud; Schmidt, Florian; Borst, Christoph; Wyatt, Jeremy L.

Proceedings of the IEEE International Conference on Intelligent Robotics and Automation: ICRA 2014. Picsatawny, NJ, USA : Institute of Electrical and Electronics Engineers (IEEE), 2014. p. 5358-5365.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Harvard

Kopicki, M, Detry, R, Schmidt, F, Borst, C & Wyatt, JL 2014, Learning dexterous grasps that generalise to novel objects by combining hand and contact models. in Proceedings of the IEEE International Conference on Intelligent Robotics and Automation: ICRA 2014. Institute of Electrical and Electronics Engineers (IEEE), Picsatawny, NJ, USA, pp. 5358-5365. https://doi.org/10.1109/ICRA.2014.6907647

APA

Kopicki, M., Detry, R., Schmidt, F., Borst, C., & Wyatt, J. L. (2014). Learning dexterous grasps that generalise to novel objects by combining hand and contact models. In Proceedings of the IEEE International Conference on Intelligent Robotics and Automation: ICRA 2014 (pp. 5358-5365). Institute of Electrical and Electronics Engineers (IEEE). https://doi.org/10.1109/ICRA.2014.6907647

Vancouver

Kopicki M, Detry R, Schmidt F, Borst C, Wyatt JL. Learning dexterous grasps that generalise to novel objects by combining hand and contact models. In Proceedings of the IEEE International Conference on Intelligent Robotics and Automation: ICRA 2014. Picsatawny, NJ, USA: Institute of Electrical and Electronics Engineers (IEEE). 2014. p. 5358-5365 https://doi.org/10.1109/ICRA.2014.6907647

Author

Kopicki, Marek ; Detry, Renaud ; Schmidt, Florian ; Borst, Christoph ; Wyatt, Jeremy L. / Learning dexterous grasps that generalise to novel objects by combining hand and contact models. Proceedings of the IEEE International Conference on Intelligent Robotics and Automation: ICRA 2014. Picsatawny, NJ, USA : Institute of Electrical and Electronics Engineers (IEEE), 2014. pp. 5358-5365

Bibtex

@inproceedings{46daf0dabafd46f19abe0e8e00703230,
title = "Learning dexterous grasps that generalise to novel objects by combining hand and contact models",
abstract = "Generalising dexterous grasps to novel objects is an open problem. We show how to learn grasps for high DoF hands that generalise to novel objects, given as little as one demonstrated grasp. During grasp learning two types of probability density are learned that model the demonstrated grasp. The first density type (the contact model) models the relationship of an individual finger part to local surface features at its contact point. The second density type (the hand configuration model) models the whole hand configuration during the approach to grasp. When presented with a new object, many candidate grasps are generated, and a kinematically feasible grasp is selected that maximises the product of these densities. We demonstrate 31 successful grasps on novel objects (an 86% success rate), transferred from 16 training grasps. The method enables: transfer of dexterous grasps within object categories; across object categories; to and from objects where there is no complete model of the object available; and using two different dexterous hands.",
author = "Marek Kopicki and Renaud Detry and Florian Schmidt and Christoph Borst and Wyatt, {Jeremy L.}",
year = "2014",
month = sep,
day = "22",
doi = "10.1109/ICRA.2014.6907647",
language = "English",
pages = "5358--5365",
booktitle = "Proceedings of the IEEE International Conference on Intelligent Robotics and Automation",
publisher = "Institute of Electrical and Electronics Engineers (IEEE)",

}

RIS

TY - GEN

T1 - Learning dexterous grasps that generalise to novel objects by combining hand and contact models

AU - Kopicki, Marek

AU - Detry, Renaud

AU - Schmidt, Florian

AU - Borst, Christoph

AU - Wyatt, Jeremy L.

PY - 2014/9/22

Y1 - 2014/9/22

N2 - Generalising dexterous grasps to novel objects is an open problem. We show how to learn grasps for high DoF hands that generalise to novel objects, given as little as one demonstrated grasp. During grasp learning two types of probability density are learned that model the demonstrated grasp. The first density type (the contact model) models the relationship of an individual finger part to local surface features at its contact point. The second density type (the hand configuration model) models the whole hand configuration during the approach to grasp. When presented with a new object, many candidate grasps are generated, and a kinematically feasible grasp is selected that maximises the product of these densities. We demonstrate 31 successful grasps on novel objects (an 86% success rate), transferred from 16 training grasps. The method enables: transfer of dexterous grasps within object categories; across object categories; to and from objects where there is no complete model of the object available; and using two different dexterous hands.

AB - Generalising dexterous grasps to novel objects is an open problem. We show how to learn grasps for high DoF hands that generalise to novel objects, given as little as one demonstrated grasp. During grasp learning two types of probability density are learned that model the demonstrated grasp. The first density type (the contact model) models the relationship of an individual finger part to local surface features at its contact point. The second density type (the hand configuration model) models the whole hand configuration during the approach to grasp. When presented with a new object, many candidate grasps are generated, and a kinematically feasible grasp is selected that maximises the product of these densities. We demonstrate 31 successful grasps on novel objects (an 86% success rate), transferred from 16 training grasps. The method enables: transfer of dexterous grasps within object categories; across object categories; to and from objects where there is no complete model of the object available; and using two different dexterous hands.

UR - http://www.scopus.com/inward/record.url?scp=84929225062&partnerID=8YFLogxK

U2 - 10.1109/ICRA.2014.6907647

DO - 10.1109/ICRA.2014.6907647

M3 - Conference contribution

AN - SCOPUS:84929225062

SP - 5358

EP - 5365

BT - Proceedings of the IEEE International Conference on Intelligent Robotics and Automation

PB - Institute of Electrical and Electronics Engineers (IEEE)

CY - Picsatawny, NJ, USA

ER -