dc.contributor.author |
Makondo, Ndivhuwo
|
|
dc.contributor.author |
Rosman, Benjamin S
|
|
dc.contributor.author |
Hasegawa, O
|
|
dc.date.accessioned |
2018-08-02T12:34:31Z |
|
dc.date.available |
2018-08-02T12:34:31Z |
|
dc.date.issued |
2018-05 |
|
dc.identifier.citation |
Makondo, N., Rosman, B.S. and Hasegawa, O. 2018. Accelerating model learning with inter-robot knowledge transfer. IEEE International Conference on Robotics and Automation (ICRA2018), 21-25 May 2018, Brisbane, Australia |
en_US |
dc.identifier.uri |
https://www.youtube.com/watch?v=P9P8eBvYxoI
|
|
dc.identifier.uri |
http://hdl.handle.net/10204/10343
|
|
dc.description |
Paper presented at the IEEE International Conference on Robotics and Automation (ICRA2018), 21-25 May 2018, Brisbane, Australia |
en_US |
dc.description.abstract |
Online learning of a robot’s inverse dynamics model for trajectory tracking necessitates an interaction between the robot and its environment to collect training data. This is challenging for physical robots in the real world, especially for humanoids and manipulators due to their large and high dimensional state and action spaces, as a large amount of data must be collected over time. This can put the robot in danger when learning tabula rasa and can also be a time-intensive process especially in a multi-robot setting, where each robot is learning its model from scratch. We propose accelerating learning of the inverse dynamics model for trajectory tracking tasks in this multi-robot setting using knowledge transfer, where robots share and re-use data collected by preexisting robots, in order to speed up learning for new robots. We propose a scheme for collecting a sample of correspondences from the robots for training transfer models, and demonstrate, in simulations, the benefit of knowledge transfer in accelerating online learning of the inverse dynamics model between several robots, including between a low-cost Interbotix PhantomX Pincher arm, and a more expensive and relatively heavier Kuka youBot arm. We show that knowledge transfer can save up to 63% of training time of the youBot arm compared to learning from scratch, and about 58% for the lighter Pincher arm. |
en_US |
dc.language.iso |
en |
en_US |
dc.relation.ispartofseries |
Worklist;20913 |
|
dc.subject |
Robot models |
en_US |
dc.subject |
Model learning |
en_US |
dc.subject |
Model transfer |
en_US |
dc.subject |
Manifold learning |
en_US |
dc.title |
Accelerating model learning with inter-robot knowledge transfer |
en_US |
dc.type |
Conference Presentation |
en_US |
dc.identifier.apacitation |
Makondo, N., Rosman, B. S., & Hasegawa, O. (2018). Accelerating model learning with inter-robot knowledge transfer. http://hdl.handle.net/10204/10343 |
en_ZA |
dc.identifier.chicagocitation |
Makondo, Ndivhuwo, Benjamin S Rosman, and O Hasegawa. "Accelerating model learning with inter-robot knowledge transfer." (2018): http://hdl.handle.net/10204/10343 |
en_ZA |
dc.identifier.vancouvercitation |
Makondo N, Rosman BS, Hasegawa O, Accelerating model learning with inter-robot knowledge transfer; 2018. http://hdl.handle.net/10204/10343 . |
en_ZA |
dc.identifier.ris |
TY - Conference Presentation
AU - Makondo, Ndivhuwo
AU - Rosman, Benjamin S
AU - Hasegawa, O
AB - Online learning of a robot’s inverse dynamics model for trajectory tracking necessitates an interaction between the robot and its environment to collect training data. This is challenging for physical robots in the real world, especially for humanoids and manipulators due to their large and high dimensional state and action spaces, as a large amount of data must be collected over time. This can put the robot in danger when learning tabula rasa and can also be a time-intensive process especially in a multi-robot setting, where each robot is learning its model from scratch. We propose accelerating learning of the inverse dynamics model for trajectory tracking tasks in this multi-robot setting using knowledge transfer, where robots share and re-use data collected by preexisting robots, in order to speed up learning for new robots. We propose a scheme for collecting a sample of correspondences from the robots for training transfer models, and demonstrate, in simulations, the benefit of knowledge transfer in accelerating online learning of the inverse dynamics model between several robots, including between a low-cost Interbotix PhantomX Pincher arm, and a more expensive and relatively heavier Kuka youBot arm. We show that knowledge transfer can save up to 63% of training time of the youBot arm compared to learning from scratch, and about 58% for the lighter Pincher arm.
DA - 2018-05
DB - ResearchSpace
DP - CSIR
KW - Robot models
KW - Model learning
KW - Model transfer
KW - Manifold learning
LK - https://researchspace.csir.co.za
PY - 2018
T1 - Accelerating model learning with inter-robot knowledge transfer
TI - Accelerating model learning with inter-robot knowledge transfer
UR - http://hdl.handle.net/10204/10343
ER -
|
en_ZA |