Generative Locomotion Model of Snake Robot with Hierarchical Networks for Topological Representation

Yoonkyu Hwang, Masato Ishikawa

Abstract

We propose novel generative locomotion models for snake robots. Locomotion researches have been relied on human experts with rich domain-knowledge and experience. Although recent data-driven approaches can achieve explict controllers to make robots move, results often do not show enough interpretability with respect to user-oriented interface. The proposed model focuses on interpretable locomotion generation to help intuitive locomotion planning by end-users. First, we introduce the topological shaping for time-series training data. This allows us to bound the data to specific region, which leads to training/inference simplification, intuitive visualization, and finally high generalization property for the proposed framework. Second, the dedicated hierarchical networks were designed to propagate complex contexts in the snake robot locomotion. The result shows that our generative locomotion models can be utilized as user-oriented interface for interpretable locomotion design.

Download


Paper Citation


in Harvard Style

Hwang Y. and Ishikawa M. (2020). Generative Locomotion Model of Snake Robot with Hierarchical Networks for Topological Representation.In Proceedings of the 12th International Conference on Agents and Artificial Intelligence - Volume 2: ICAART, ISBN 978-989-758-395-7, pages 194-202. DOI: 10.5220/0008981401940202


in Bibtex Style

@conference{icaart20,
author={Yoonkyu Hwang and Masato Ishikawa},
title={Generative Locomotion Model of Snake Robot with Hierarchical Networks for Topological Representation},
booktitle={Proceedings of the 12th International Conference on Agents and Artificial Intelligence - Volume 2: ICAART,},
year={2020},
pages={194-202},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0008981401940202},
isbn={978-989-758-395-7},
}


in EndNote Style

TY - CONF

JO - Proceedings of the 12th International Conference on Agents and Artificial Intelligence - Volume 2: ICAART,
TI - Generative Locomotion Model of Snake Robot with Hierarchical Networks for Topological Representation
SN - 978-989-758-395-7
AU - Hwang Y.
AU - Ishikawa M.
PY - 2020
SP - 194
EP - 202
DO - 10.5220/0008981401940202