JZUS - Journal of Zhejiang University SCIENCE

Journal of Zhejiang University SCIENCE A

Accepted manuscript available online (unedited version)

Stable and continuous vertical jumping control of hydraulic legged robots through reinforcement learning

Author(s): Junhui ZHANG, Pengyuan JI, Lizhou FANG, Jinyuan LIU, Dandan WANG, Jikun AI, Huaizhi ZONG, Bing XU
Affiliation(s): State Key Laboratory of Fluid Power and Mechatronic Systems, Zhejiang University, Hangzhou 310058, China
Corresponding email(s): przhao@163.com
Key Words: Legged robot; Deep reinforcement learning; Quasi-realistic modelling; Hydraulic system; Jumping control

Share this article to： More <<< Previous Paper \|Next Paper >>>

Junhui ZHANG, Pengyuan JI, Lizhou FANG, Jinyuan LIU, Dandan WANG, Jikun AI, Huaizhi ZONG, Bing XU. Stable and continuous vertical jumping control of hydraulic legged robots through reinforcement learning[J]. Journal of Zhejiang University Science A,in press.Frontiers of Information Technology & Electronic Engineering,in press.https://doi.org/10.1631/jzus.A2500142

@article{title="Stable and continuous vertical jumping control of hydraulic legged robots through reinforcement learning",
author="Junhui ZHANG, Pengyuan JI, Lizhou FANG, Jinyuan LIU, Dandan WANG, Jikun AI, Huaizhi ZONG, Bing XU",
journal="Journal of Zhejiang University Science A",
year="in press",
publisher="Zhejiang University Press & Springer",
doi="https://doi.org/10.1631/jzus.A2500142"
}

%0 Journal Article
%T Stable and continuous vertical jumping control of hydraulic legged robots through reinforcement learning
%A Junhui ZHANG
%A Pengyuan JI
%A Lizhou FANG
%A Jinyuan LIU
%A Dandan WANG
%A Jikun AI
%A Huaizhi ZONG
%A Bing XU
%J Journal of Zhejiang University SCIENCE A
%P
%@ 1673-565X
%D in press
%I Zhejiang University Press & Springer
doi="https://doi.org/10.1631/jzus.A2500142"

TY - JOUR
T1 - Stable and continuous vertical jumping control of hydraulic legged robots through reinforcement learning
A1 - Junhui ZHANG
A1 - Pengyuan JI
A1 - Lizhou FANG
A1 - Jinyuan LIU
A1 - Dandan WANG
A1 - Jikun AI
A1 - Huaizhi ZONG
A1 - Bing XU
J0 - Journal of Zhejiang University Science A
SP -
EP -
%@ 1673-565X
Y1 - in press
PB - Zhejiang University Press & Springer
ER -
doi="https://doi.org/10.1631/jzus.A2500142"

Abstract
Chinese Summary
Academic Network
Reviewer Comment

Abstract: Hydraulic legged robots have potential for highly dynamic motion due to their large power-to-weight ratios. However, it is challenging to ensure both stability and continuity in the motion of such robots. In this study, we propose a jumping motion control framework based on deep reinforcement learning that enables hydraulic limb leg units to perform stable and continuous jumping motions. First, to accurately represent the performance of a physical prototype, a quasi-realistic model incorporating physical feasibility constraints is constructed. This model is informed by analysis of the relevant fluid dynamics, and incorporates a trajectory generator and a motion tracking controller. To achieve stable and continuous jumping performance, a deep reinforcement learning algorithm is developed which jointly optimizes the trajectory generator and the motion tracking controller. Through validation on the physical prototype, we demonstrate that the proposed method reduces the maximum deviation and the average deviation by over 47% and 60%, respectively, and improves landing compliance by up to 7.7% compared to a baseline optimization algorithm, the NSGA-II. The proposed control framework may serve as a reference for high dynamic motion control of legged robots, and multi-objective optimization across several decision variables.

Darkslateblue:Affiliate; Royal Blue:Author; Turquoise:Article

Reference

Open peer comments: Debate/Discuss/Question/Opinion

<1>

- Go to

Darkslateblue:Affiliate; Royal Blue:Author; Turquoise:Article

Reference