The Locality of Data-Driven Models

B Bittner, Shai Revzen

Dynamic Walking, 2018

Abstract

Data-driven modeling of gaits allows for the fast generation of predictive motion models from experimental data. An important consideration is the locality of these models. In what state space of the robot do data-driven models maintain predictive quality? We present our approach to measuring this notion of locality.

BibTeX

@conference{bittner-2018-dw-lddm,
  author = {Bittner, B and Revzen, Shai},
  title = {The Locality of Data-Driven Models},
  booktitle = {Dynamic Walking},
  year = {2018}
}