Demonstration-guided motion planning

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

Abstract

We present demonstration-guided motion planning (DGMP), a new frame-work for planning motions for personal robots to perform household tasks. DGMP combines the strengths of sampling-based motion planning and robot learning from demonstrations to generate plans that (1) avoid novel obstacles in cluttered environments, and (2) learn and maintain critical aspects of the motion required to successfully accomplish a task. Sampling-based motion planning methods are highly effective at computing paths from start to goal configurations that avoid obstacles, but task constraints (e.g. a glass of water must be held upright to avoid a spill) must be explicitly enumerated and programmed. Instead, we use a set of expert demonstrations and automatically extract time-dependent task constraints by learning low variance aspects of the demonstrations, which are correlated with the task constraints. We then introduce multi-component rapidlyexploring roadmaps (MC-RRM), a sampling-based method that incrementally computes a motion plan that avoids obstacles and optimizes a learned cost metric. We demonstrate the effectiveness of DGMP using the Aldebaran Nao robot performing household tasks in a cluttered environment, including moving a spoon full of sugar from a bowl to a cup and cleaning the surface of a table.

LanguageEnglish (US)
Title of host publicationRobotics Research - The 15th International Symposium ISRR
PublisherSpringer Verlag
Pages291-307
Number of pages17
Volume100
ISBN (Print)9783319293622
DOIs
StatePublished - Jan 1 2017
Event15th International Symposium of Robotics Research, 2011 - Flagstaff, United States
Duration: Dec 9 2011Dec 12 2011

Publication series

NameSpringer Tracts in Advanced Robotics
Volume100
ISSN (Print)1610-7438
ISSN (Electronic)1610-742X

Other

Other15th International Symposium of Robotics Research, 2011
CountryUnited States
CityFlagstaff
Period12/9/1112/12/11

Fingerprint

Motion planning
Demonstrations
Sampling
Robot learning
Robots
Hazardous materials spills
Sugars
Cleaning
Glass
Costs
Water

ASJC Scopus subject areas

  • Artificial Intelligence
  • Electrical and Electronic Engineering

Cite this

Ye, G., & Alterovitz, R. (2017). Demonstration-guided motion planning. In Robotics Research - The 15th International Symposium ISRR (Vol. 100, pp. 291-307). (Springer Tracts in Advanced Robotics; Vol. 100). Springer Verlag. DOI: 10.1007/978-3-319-29363-9_17

Demonstration-guided motion planning. / Ye, Gu; Alterovitz, Ron.

Robotics Research - The 15th International Symposium ISRR. Vol. 100 Springer Verlag, 2017. p. 291-307 (Springer Tracts in Advanced Robotics; Vol. 100).

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

Ye, G & Alterovitz, R 2017, Demonstration-guided motion planning. in Robotics Research - The 15th International Symposium ISRR. vol. 100, Springer Tracts in Advanced Robotics, vol. 100, Springer Verlag, pp. 291-307, 15th International Symposium of Robotics Research, 2011, Flagstaff, United States, 12/9/11. DOI: 10.1007/978-3-319-29363-9_17
Ye G, Alterovitz R. Demonstration-guided motion planning. In Robotics Research - The 15th International Symposium ISRR. Vol. 100. Springer Verlag. 2017. p. 291-307. (Springer Tracts in Advanced Robotics). Available from, DOI: 10.1007/978-3-319-29363-9_17
Ye, Gu ; Alterovitz, Ron. / Demonstration-guided motion planning. Robotics Research - The 15th International Symposium ISRR. Vol. 100 Springer Verlag, 2017. pp. 291-307 (Springer Tracts in Advanced Robotics).
@inproceedings{2c2e90ff8f204cadabff731fd473c1c9,
title = "Demonstration-guided motion planning",
abstract = "We present demonstration-guided motion planning (DGMP), a new frame-work for planning motions for personal robots to perform household tasks. DGMP combines the strengths of sampling-based motion planning and robot learning from demonstrations to generate plans that (1) avoid novel obstacles in cluttered environments, and (2) learn and maintain critical aspects of the motion required to successfully accomplish a task. Sampling-based motion planning methods are highly effective at computing paths from start to goal configurations that avoid obstacles, but task constraints (e.g. a glass of water must be held upright to avoid a spill) must be explicitly enumerated and programmed. Instead, we use a set of expert demonstrations and automatically extract time-dependent task constraints by learning low variance aspects of the demonstrations, which are correlated with the task constraints. We then introduce multi-component rapidlyexploring roadmaps (MC-RRM), a sampling-based method that incrementally computes a motion plan that avoids obstacles and optimizes a learned cost metric. We demonstrate the effectiveness of DGMP using the Aldebaran Nao robot performing household tasks in a cluttered environment, including moving a spoon full of sugar from a bowl to a cup and cleaning the surface of a table.",
author = "Gu Ye and Ron Alterovitz",
year = "2017",
month = "1",
day = "1",
doi = "10.1007/978-3-319-29363-9_17",
language = "English (US)",
isbn = "9783319293622",
volume = "100",
series = "Springer Tracts in Advanced Robotics",
publisher = "Springer Verlag",
pages = "291--307",
booktitle = "Robotics Research - The 15th International Symposium ISRR",
address = "Germany",

}

TY - GEN

T1 - Demonstration-guided motion planning

AU - Ye,Gu

AU - Alterovitz,Ron

PY - 2017/1/1

Y1 - 2017/1/1

N2 - We present demonstration-guided motion planning (DGMP), a new frame-work for planning motions for personal robots to perform household tasks. DGMP combines the strengths of sampling-based motion planning and robot learning from demonstrations to generate plans that (1) avoid novel obstacles in cluttered environments, and (2) learn and maintain critical aspects of the motion required to successfully accomplish a task. Sampling-based motion planning methods are highly effective at computing paths from start to goal configurations that avoid obstacles, but task constraints (e.g. a glass of water must be held upright to avoid a spill) must be explicitly enumerated and programmed. Instead, we use a set of expert demonstrations and automatically extract time-dependent task constraints by learning low variance aspects of the demonstrations, which are correlated with the task constraints. We then introduce multi-component rapidlyexploring roadmaps (MC-RRM), a sampling-based method that incrementally computes a motion plan that avoids obstacles and optimizes a learned cost metric. We demonstrate the effectiveness of DGMP using the Aldebaran Nao robot performing household tasks in a cluttered environment, including moving a spoon full of sugar from a bowl to a cup and cleaning the surface of a table.

AB - We present demonstration-guided motion planning (DGMP), a new frame-work for planning motions for personal robots to perform household tasks. DGMP combines the strengths of sampling-based motion planning and robot learning from demonstrations to generate plans that (1) avoid novel obstacles in cluttered environments, and (2) learn and maintain critical aspects of the motion required to successfully accomplish a task. Sampling-based motion planning methods are highly effective at computing paths from start to goal configurations that avoid obstacles, but task constraints (e.g. a glass of water must be held upright to avoid a spill) must be explicitly enumerated and programmed. Instead, we use a set of expert demonstrations and automatically extract time-dependent task constraints by learning low variance aspects of the demonstrations, which are correlated with the task constraints. We then introduce multi-component rapidlyexploring roadmaps (MC-RRM), a sampling-based method that incrementally computes a motion plan that avoids obstacles and optimizes a learned cost metric. We demonstrate the effectiveness of DGMP using the Aldebaran Nao robot performing household tasks in a cluttered environment, including moving a spoon full of sugar from a bowl to a cup and cleaning the surface of a table.

UR - http://www.scopus.com/inward/record.url?scp=84984817815&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84984817815&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-29363-9_17

DO - 10.1007/978-3-319-29363-9_17

M3 - Conference contribution

SN - 9783319293622

VL - 100

T3 - Springer Tracts in Advanced Robotics

SP - 291

EP - 307

BT - Robotics Research - The 15th International Symposium ISRR

PB - Springer Verlag

ER -