Differentiable Physics and Stable Modes for Tool-Use and Manipulation Planning
Name
72d26e5093576b4de943b8310232d04ea54d.pdf
Description
Accepted version
Size
1.9 MB
Format
Adobe PDF
Checksum (MD5)
078464401af7a86d74e09817bfb5c18e
Author(s)
Toussaint, Marc A
Allen, Kelsey Rebecca
Smith, Kevin A
Tenenbaum, Joshua B
Date Issued
June 2018
Journal
Robotics: Science and Systems 2018
Publisher
Robotics: Science and Systems Foundation
Citation
Toussaint, Marc et al. “Differentiable Physics and Stable Modes for Tool-Use and Manipulation Planning.” Proceedings of the Robotics: Science and Systems, RSS 2018, Pittsburgh, Pennsylvania, June 26-30, 2018 © 2018 The Author(s)
Version
Author's final manuscript
Abstract
We consider the problem of sequential manipulationand tool-use planning in domains that include physical interac-tions such as hitting and throwing. The approach integrates aTask And Motion Planning formulation with primitives that ei-ther impose stable kinematic constraints or differentiable dynam-ical and impulse exchange constraints at the path optimizationlevel. We demonstrate our approach on a variety of physicalpuzzles that involve tool use and dynamic interactions. We thencompare manipulation sequences generated by our approach tohuman actions on analogous tasks, suggesting future directionsand illuminating current limitations.
Subjects
German Research Foundation (Grant TO 409/9-1)
MIT Department
Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory
Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences
Terms of Use
Creative Commons Attribution-Noncommercial-Share Alike
Persistent DSpace Link
DOI of Published Version
https://doi.org/10.15607/RSS.2018.XIV.044
