Augmenting Physical Simulators with Stochastic Neural Networks: Case Study of Planar Pushing and Bouncing
Name
ajay-iros18.pdf
Description
Accepted version
Size
866.03 KB
Format
Adobe PDF
Checksum (MD5)
fac4fee791d926be8652d3b205578e43
Author(s)
Ajay, Anurag
Wu, Jiajun
Fazeli, Nima
Bauza, Maria
Kaelbling, Leslie P.
Tenenbaum, Joshua B.
Rodriguez, Alberto
Date Issued
October 2018
Publisher
IEEE
Citation
Ajay, Anurag, Wu, Jiajun, Fazeli, Nima, Bauza, Maria, Kaelbling, Leslie P. et al. 2018. "Augmenting Physical Simulators with Stochastic Neural Networks: Case Study of Planar Pushing and Bouncing."
Version
Author's final manuscript
Abstract
© 2018 IEEE. An efficient, generalizable physical simulator with universal uncertainty estimates has wide applications in robot state estimation, planning, and control. In this paper, we build such a simulator for two scenarios, planar pushing and ball bouncing, by augmenting an analytical rigid-body simulator with a neural network that learns to model uncertainty as residuals. Combining symbolic, deterministic simulators with learnable, stochastic neural nets provides us with expressiveness, efficiency, and generalizability simultaneously. Our model outperforms both purely analytical and purely learned simulators consistently on real, standard benchmarks. Compared with methods that model uncertainty using Gaussian processes, our model runs much faster, generalizes better to new object shapes, and is able to characterize the complex distribution of object trajectories.
MIT Department
Massachusetts Institute of Technology. Department of Mechanical Engineering
Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory
Terms of Use
Creative Commons Attribution-Noncommercial-Share Alike
Persistent DSpace Link
DOI of Published Version
https://doi.org/10.1109/iros.2018.8593995
