Generalizing Behavior Trees and Motion-Generator (BTMG) Policy Representation for Robotic Tasks Over Scenario Parameters
Ahmad, Faseeh LU ; Mayr, Matthias LU
; Topp, Elin Anna
LU
; Malec, Jacek
LU
and Krueger, Volker
LU
(2022)
- Abstract
- We propose a generalisation of a behaviour tree and motiongenerator based robot arm policy representation for learning and solving tasks such as contact-rich tasks like peg insertion or pushing an object. We use planning to generate skill sequences needed to execute these tasks and rely on reinforcement learning to obtain parameters of the policy. We assume gaussian processes as a suitable method for this generalisation and present preliminary, promising results from initial experiments.
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/aed2dfb2-0343-4655-9f25-29852c81eb70
- author
- Ahmad, Faseeh
LU
; Mayr, Matthias
LU
; Topp, Elin Anna
LU
; Malec, Jacek
LU
and Krueger, Volker
LU
- organization
- publishing date
- 2022
- type
- Contribution to conference
- publication status
- published
- subject
- keywords
- Generalization, Industrial Robots, Reinforcement learning
- pages
- 2 pages
- project
- Robotics and Semantic Systems
- WASP Professor Package: Cognitive Robots for Manufacturing
- RobotLab LTH
- Efficient Learning of Robot Skills
- language
- English
- LU publication?
- yes
- id
- aed2dfb2-0343-4655-9f25-29852c81eb70
- alternative location
- https://prl-theworkshop.github.io/prl2022-ijcai/papers/PRL2022_paper_25.pdf
- date added to LUP
- 2022-09-19 20:15:26
- date last changed
- 2024-02-05 14:41:47
@misc{aed2dfb2-0343-4655-9f25-29852c81eb70,
abstract = {{We propose a generalisation of a behaviour tree and motiongenerator based robot arm policy representation for learning and solving tasks such as contact-rich tasks like peg insertion or pushing an object. We use planning to generate skill sequences needed to execute these tasks and rely on reinforcement learning to obtain parameters of the policy. We assume gaussian processes as a suitable method for this generalisation and present preliminary, promising results from initial experiments.}},
author = {{Ahmad, Faseeh and Mayr, Matthias and Topp, Elin Anna and Malec, Jacek and Krueger, Volker}},
keywords = {{Generalization; Industrial Robots; Reinforcement learning}},
language = {{eng}},
title = {{Generalizing Behavior Trees and Motion-Generator (BTMG) Policy Representation for Robotic Tasks Over Scenario Parameters}},
url = {{https://prl-theworkshop.github.io/prl2022-ijcai/papers/PRL2022_paper_25.pdf}},
year = {{2022}},
}