Contact-History-Based Velocity-Tracking Reward Derived from Center-of-Mass Controllability for Quadruped Locomotion
Robotics · 07.11.2026 · 15 (11) · 231
DOI: 10.3390/robotics15110231 ↗Kurz gefasst
Fachartikel in der Zeitschrift Robotics, online erschienen am 07.11.2026. Verfasst von Kai Ito und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Japan Society for the Promotion of Science.
Abstract (Originalsprache)
Velocity tracking is a fundamental task in reinforcement learning for quadruped locomotion. Conventional velocity-tracking rewards are typically formulated in terms of the squared error between commanded and actual velocities. In this study, we propose a velocity-tracking reward from the controllability Gramian of center-of-mass (CoM) dynamics, with ground reaction forces (GRFs) as control inputs. The proposed reward dynamically scales the velocity error according to recent contact history. Although derived from the controllability Gramian of CoM dynamics with GRFs as control inputs, the resulting weighting factor depends only on the number of feet in contact over a recent time window, without dependence on body rotation or the linearized dynamics. We compared policies trained using the proposed reward with those trained using a conventional squared-error-based velocity-tracking reward. When used as the translational velocity-tracking term, the proposed reward reduced the cost of transport (CoT) by approximately 29% and the time-averaged squared vertical velocity by approximately 93.4%. However, the CoT improvement was not observed when the energy reward was removed, indicating that the improvement depends on the combination of the proposed reward and the energy reward. In addition, increasing the weight of the proposed reward tended to shift the learned gait from bound to trot. These results suggest that incorporating CoM controllability into velocity tracking suppresses unnecessary vertical motion and promotes the learning of body states suitable for forward locomotion.
Förderung
- Japan Society for the Promotion of Science
Zitieren
APA
Ito, K., & Ikemoto, Y. (2026). Contact-History-Based Velocity-Tracking Reward Derived from Center-of-Mass Controllability for Quadruped Locomotion. Robotics, 15(11), 231. https://doi.org/10.3390/robotics15110231
Vancouver
Ito K, Ikemoto Y. Contact-History-Based Velocity-Tracking Reward Derived from Center-of-Mass Controllability for Quadruped Locomotion. Robotics. 2026;15(11):231. doi:10.3390/robotics15110231
BibTeX
@article{Ito2026,
author = {Ito, Kai and Ikemoto, Yusuke},
title = {Contact-History-Based Velocity-Tracking Reward Derived from Center-of-Mass Controllability for Quadruped Locomotion},
journal = {Robotics},
year = {2026},
volume = {15},
number = {11},
pages = {231},
doi = {10.3390/robotics15110231}
}
EUVIMED