Constrained Reinforcement Learning for Unstable Point-Feet Bipedal Locomotion Applied to the Bolt Robot

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Roux, Constant, Chane-Sane, Elliot, De Matteïs, Ludovic, Flayols, Thomas, Manhes, Jérôme, Stasse, Olivier, Souères, Philippe
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909720626855936
author Roux, Constant
Chane-Sane, Elliot
De Matteïs, Ludovic
Flayols, Thomas
Manhes, Jérôme
Stasse, Olivier
Souères, Philippe
author_facet Roux, Constant
Chane-Sane, Elliot
De Matteïs, Ludovic
Flayols, Thomas
Manhes, Jérôme
Stasse, Olivier
Souères, Philippe
contents Bipedal locomotion is a key challenge in robotics, particularly for robots like Bolt, which have a point-foot design. This study explores the control of such underactuated robots using constrained reinforcement learning, addressing their inherent instability, lack of arms, and limited foot actuation. We present a methodology that leverages Constraints-as-Terminations and domain randomization techniques to enable sim-to-real transfer. Through a series of qualitative and quantitative experiments, we evaluate our approach in terms of balance maintenance, velocity control, and responses to slip and push disturbances. Additionally, we analyze autonomy through metrics like the cost of transport and ground reaction force. Our method advances robust control strategies for point-foot bipedal robots, offering insights into broader locomotion.
format Preprint
id arxiv_https___arxiv_org_abs_2508_02194
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Constrained Reinforcement Learning for Unstable Point-Feet Bipedal Locomotion Applied to the Bolt Robot
Roux, Constant
Chane-Sane, Elliot
De Matteïs, Ludovic
Flayols, Thomas
Manhes, Jérôme
Stasse, Olivier
Souères, Philippe
Robotics
Bipedal locomotion is a key challenge in robotics, particularly for robots like Bolt, which have a point-foot design. This study explores the control of such underactuated robots using constrained reinforcement learning, addressing their inherent instability, lack of arms, and limited foot actuation. We present a methodology that leverages Constraints-as-Terminations and domain randomization techniques to enable sim-to-real transfer. Through a series of qualitative and quantitative experiments, we evaluate our approach in terms of balance maintenance, velocity control, and responses to slip and push disturbances. Additionally, we analyze autonomy through metrics like the cost of transport and ground reaction force. Our method advances robust control strategies for point-foot bipedal robots, offering insights into broader locomotion.
title Constrained Reinforcement Learning for Unstable Point-Feet Bipedal Locomotion Applied to the Bolt Robot
topic Robotics
url https://arxiv.org/abs/2508.02194