IRUS Total

Learning to exploit passive compliance for energy-efficient gait generation on a compliant humanoid

File Description SizeFormat 
Kormushev_AURO-2018[1].pdfAccepted version2.25 MBAdobe PDFView/Open
Title: Learning to exploit passive compliance for energy-efficient gait generation on a compliant humanoid
Authors: Kormushev, P
Ugurlu, B
Caldwell, DG
Tsagarakis, NG
Item Type: Journal Article
Abstract: Modern humanoid robots include not only active compliance but also passive compliance. Apart from improved safety and dependability, availability of passive elements, such as springs, opens up new possibilities for improving the energy efficiency. With this in mind, this paper addresses the challenging open problem of exploiting the passive compliance for the purpose of energy efficient humanoid walking. To this end, we develop a method comprising two parts: an optimization part that finds an optimal vertical center-of-mass trajectory, and a walking pattern generator part that uses this trajectory to produce a dynamically-balanced gait. For the optimization part, we propose a reinforcement learning approach that dynamically evolves the policy parametrization during the learning process. By gradually increasing the representational power of the policy parametrization, it manages to find better policies in a faster and computationally efficient way. For the walking generator part, we develop a variable-center-of-mass-height ZMP-based bipedal walking pattern generator. The method is tested in real-world experiments with the bipedal robot COMAN and achieves a significant 18% reduction in the electric energy consumption by learning to efficiently use the passive compliance of the robot.
Issue Date: 1-Jan-2019
Date of Acceptance: 12-Jan-2018
URI: http://hdl.handle.net/10044/1/60782
DOI: https://dx.doi.org/10.1007/s10514-018-9697-6
ISSN: 1573-7527
Publisher: Springer
Start Page: 79
End Page: 95
Journal / Book Title: Autonomous Robots
Volume: 43
Issue: 1
Copyright Statement: © 2018 Springer-Verlag. The final publication is available at Springer via https://dx.doi.org/10.1007/s10514-018-9697-6
Keywords: Science & Technology
Computer Science, Artificial Intelligence
Computer Science
Bipedal walking
Energy efficiency
Reinforcement learning
Passive compliance
Industrial Engineering & Automation
0801 Artificial Intelligence and Image Processing
1702 Cognitive Sciences
0913 Mechanical Engineering
Publication Status: Published
Online Publication Date: 2018-02-13
Appears in Collections:Faculty of Engineering
Dyson School of Design Engineering