A Real-Robot Dataset for Assessing Transferability of Learned Dynamics Models

DOI

In the context of model-based reinforcement learning and control, a large number of methods for learning system dynamics have been proposed in recent years. The purpose of these learned models is to synthesize new control policies. An important open question is how robust current dynamics-learning methods are to shifts in the data distribution due to changes in the control policy. We present a real-robot dataset which allows to systematically investigate this question. This dataset contains trajectories of a 3 degrees-of-freedom (DOF) robot being controlled by a diverse set of policies. Software to reproduce our benchmark of a few widely-used dynamics-learning methods using the proposed dataset is available in our code repository

Identifier
DOI https://doi.org/10.17617/3.ZT6K7P
Metadata Access https://edmond.mpg.de/api/datasets/export?exporter=dataverse_json&persistentId=doi:10.17617/3.ZT6K7P
Provenance
Creator Agudelo-España, Diego; Zadaianchuk, Andrii; Wenk, Philipp; Garg, Aditya; Akpo, Joel Bessekon; Grimminger, Felix; Viereck, Julian; Naveau, Maximilien; Righetti, Ludovic ‬; Martius, Georg; Krause, Andreas; Schölkopf, Bernhard; Bauer, Stefan; Wüthrich, Manuel
Publisher Edmond
Publication Year 2022
OpenAccess true
Contact DIEGO.AGUDELO(at)TUEBINGEN.MPG.DE
Representation
Language English
Resource Type Dataset
Version 1
Discipline Other