Sim-to-Real Transfer with Incremental Environment Complexity for Reinforcement Learning of Depth-based Robot Navigation

Thomas Chaffre, Julien Moras, Adrien Chan-Hon-Tong, Julien Marzat

2020

Abstract

Transferring learning-based models to the real world remains one of the hardest problems in model-free control theory. Due to the cost of data collection on a real robot and the limited sample efficiency of Deep Reinforcement Learning algorithms, models are usually trained in a simulator which theoretically provides an infinite amount of data. Despite offering unbounded trial and error runs, the reality gap between simulation and the physical world brings little guarantee about the policy behavior in real operation. Depending on the problem, expensive real fine-tuning and/or a complex domain randomization strategy may be required to produce a relevant policy. In this paper, a Soft-Actor Critic (SAC) training strategy using incremental environment complexity is proposed to drastically reduce the need for additional training in the real world. The application addressed is depth-based mapless navigation, where a mobile robot should reach a given waypoint in a cluttered environment with no prior mapping information. Experimental results in simulated and real environments are presented to assess quantitatively the efficiency of the proposed approach, which demonstrated a success rate twice higher than a naive strategy.

Paper Citation

in Harvard Style

Chaffre T., Moras J., Chan-Hon-Tong A. and Marzat J. (2020). Sim-to-Real Transfer with Incremental Environment Complexity for Reinforcement Learning of Depth-based Robot Navigation.In Proceedings of the 17th International Conference on Informatics in Control, Automation and Robotics - Volume 1: ICINCO, ISBN 978-989-758-442-8, pages 314-323. DOI: 10.5220/0009821603140323

in Bibtex Style

@conference{icinco20,
author={Thomas Chaffre and Julien Moras and Adrien Chan-Hon-Tong and Julien Marzat},
title={Sim-to-Real Transfer with Incremental Environment Complexity for Reinforcement Learning of Depth-based Robot Navigation},
booktitle={Proceedings of the 17th International Conference on Informatics in Control, Automation and Robotics - Volume 1: ICINCO,},
year={2020},
pages={314-323},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0009821603140323},
isbn={978-989-758-442-8},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 17th International Conference on Informatics in Control, Automation and Robotics - Volume 1: ICINCO,
TI - Sim-to-Real Transfer with Incremental Environment Complexity for Reinforcement Learning of Depth-based Robot Navigation
SN - 978-989-758-442-8
AU - Chaffre T.
AU - Moras J.
AU - Chan-Hon-Tong A.
AU - Marzat J.
PY - 2020
SP - 314
EP - 323
DO - 10.5220/0009821603140323