A modular approach to learning manipulation strategies from human demonstration

Bidan Huang, Miao Li, Ravin Luis De Souza, Joanna J. Bryson, Aude Billard

Research output: Contribution to journalArticle

  • 5 Citations

Abstract

Object manipulation is a challenging task for robotics, as the physics involved in object interaction is complex and hard to express analytically. Here we introduce a modular approach for learning a manipulation strategy from human demonstration. Firstly we record a human performing a task that requires an adaptive control strategy in different conditions, i.e. different task contexts. We then perform modular decomposition of the control strategy, using phases of the recorded actions to guide segmentation. Each module represents a part of the strategy, encoded as a pair of forward and inverse models. All modules contribute to the final control policy; their recommendations are integrated via a system of weighting based on their own estimated error in the current task context. We validate our approach by demonstrating it, both in a simulation for clarity, and on a real robot platform to demonstrate robustness and capacity to generalise. The robot task is opening bottle caps. We show that our approach can modularize an adaptive control strategy and generate appropriate motor commands for the robot to accomplish the complete task, even for novel bottles.
LanguageEnglish
Pages1-25
Number of pages25
JournalAutonomous Robots
Volume40
Issue number5
Early online date5 Oct 2015
DOIs
StatusPublished - 2015

Fingerprint

Demonstrations
Bottles
Robots
Container closures
Robotics
Physics
Decomposition

Cite this

A modular approach to learning manipulation strategies from human demonstration. / Huang, Bidan; Li, Miao; De Souza, Ravin Luis; Bryson, Joanna J.; Billard, Aude.

In: Autonomous Robots, Vol. 40, No. 5, 2015, p. 1-25.

Research output: Contribution to journalArticle

Huang, Bidan ; Li, Miao ; De Souza, Ravin Luis ; Bryson, Joanna J. ; Billard, Aude. / A modular approach to learning manipulation strategies from human demonstration. In: Autonomous Robots. 2015 ; Vol. 40, No. 5. pp. 1-25
@article{6acac5cb9dbe4c4986346f448dad7159,
title = "A modular approach to learning manipulation strategies from human demonstration",
abstract = "Object manipulation is a challenging task for robotics, as the physics involved in object interaction is complex and hard to express analytically. Here we introduce a modular approach for learning a manipulation strategy from human demonstration. Firstly we record a human performing a task that requires an adaptive control strategy in different conditions, i.e. different task contexts. We then perform modular decomposition of the control strategy, using phases of the recorded actions to guide segmentation. Each module represents a part of the strategy, encoded as a pair of forward and inverse models. All modules contribute to the final control policy; their recommendations are integrated via a system of weighting based on their own estimated error in the current task context. We validate our approach by demonstrating it, both in a simulation for clarity, and on a real robot platform to demonstrate robustness and capacity to generalise. The robot task is opening bottle caps. We show that our approach can modularize an adaptive control strategy and generate appropriate motor commands for the robot to accomplish the complete task, even for novel bottles.",
author = "Bidan Huang and Miao Li and {De Souza}, {Ravin Luis} and Bryson, {Joanna J.} and Aude Billard",
year = "2015",
doi = "10.1007/s10514-015-9501-9",
language = "English",
volume = "40",
pages = "1--25",
journal = "Autonomous Robots",
issn = "0929-5593",
publisher = "Springer Netherlands",
number = "5",

}

TY - JOUR

T1 - A modular approach to learning manipulation strategies from human demonstration

AU - Huang,Bidan

AU - Li,Miao

AU - De Souza,Ravin Luis

AU - Bryson,Joanna J.

AU - Billard,Aude

PY - 2015

Y1 - 2015

N2 - Object manipulation is a challenging task for robotics, as the physics involved in object interaction is complex and hard to express analytically. Here we introduce a modular approach for learning a manipulation strategy from human demonstration. Firstly we record a human performing a task that requires an adaptive control strategy in different conditions, i.e. different task contexts. We then perform modular decomposition of the control strategy, using phases of the recorded actions to guide segmentation. Each module represents a part of the strategy, encoded as a pair of forward and inverse models. All modules contribute to the final control policy; their recommendations are integrated via a system of weighting based on their own estimated error in the current task context. We validate our approach by demonstrating it, both in a simulation for clarity, and on a real robot platform to demonstrate robustness and capacity to generalise. The robot task is opening bottle caps. We show that our approach can modularize an adaptive control strategy and generate appropriate motor commands for the robot to accomplish the complete task, even for novel bottles.

AB - Object manipulation is a challenging task for robotics, as the physics involved in object interaction is complex and hard to express analytically. Here we introduce a modular approach for learning a manipulation strategy from human demonstration. Firstly we record a human performing a task that requires an adaptive control strategy in different conditions, i.e. different task contexts. We then perform modular decomposition of the control strategy, using phases of the recorded actions to guide segmentation. Each module represents a part of the strategy, encoded as a pair of forward and inverse models. All modules contribute to the final control policy; their recommendations are integrated via a system of weighting based on their own estimated error in the current task context. We validate our approach by demonstrating it, both in a simulation for clarity, and on a real robot platform to demonstrate robustness and capacity to generalise. The robot task is opening bottle caps. We show that our approach can modularize an adaptive control strategy and generate appropriate motor commands for the robot to accomplish the complete task, even for novel bottles.

UR - http://dx.doi.org/10.1007/s10514-015-9501-9

U2 - 10.1007/s10514-015-9501-9

DO - 10.1007/s10514-015-9501-9

M3 - Article

VL - 40

SP - 1

EP - 25

JO - Autonomous Robots

T2 - Autonomous Robots

JF - Autonomous Robots

SN - 0929-5593

IS - 5

ER -