What question did this study set out to answer?

The aim is to implement a method for generating walking policies in humanoid robots using reinforcement learning.

March 26, 2026The Proceedings of Mechanical Engineering Congress Japan0 citationsOpen Access

An Implementation of A Reinforcement Learning-Assisted Robot Task Motion Generation Algorithm

MFMasaki FUJIUE MFMasaki FUKATA TSTakeshi SATO

Key Points

The aim is to implement a method for generating walking policies in humanoid robots using reinforcement learning.
Developed a reinforcement learning-assisted algorithm for walking tasks.
Integrated the walking policy into the HRP-2 robot's control system.
Conducted walking simulations and experiments to test the implementation.
The robot successfully performed mobility tasks during simulations.
Demonstrated improved task performance compared to previous methods.

Abstract

In this paper, a method is proposed for implementing walking policies for HRP-2 humanoid robot using the Genesis. The implemented policy is integrated into the control system of HRP-2 to conduct walking simulations and experiments. This approach is aimed at enabling the robot to conduct mobility tasks.

AI से पूछें

Bookmark

View Full Paper