In-Flight Attitude Control of a Quadruped using Deep Reinforcement Learning

Tarek El-Agroudi, Finn Gross Maurer, Jørgen Anker Olsen, Kostas Alexis

Keywords: Deep Reinforcement Learning, Legged Robotics

TL;DR: We use DRL to train and demonstrate an in-flight attitude control law for a small low-cost quadruped with a five-bar-linkage leg design using only its legs as reaction masses.

Abstract: We present the development and real world demonstration of an in-flight attitude control law for a small low-cost quadruped with a five-bar-linkage leg design using only its legs as reaction masses. The control law is trained using deep reinforcement learning (DRL) and specifically through Proximal Policy Optimization (PPO) in the NVIDIA Omniverse Isaac Sim simulator with a GPU-accelerated DRL pipeline. To demonstrate the policy, a small quadruped is designed, constructed, and evaluated both on a rotating pole test setup and in free fall. During a free fall of 0.7 seconds, the quadruped follows commanded attitude steps of 45 degrees in all principal axes, and achieves an average base angular velocity of 110 degrees per second during large attitude reference steps.

Video: https://www.youtube.com/watch?v=5qNPCH34M2M&t=1s

Website: https://finnfi.github.io/

Code: https://github.com/ntnu-arl/Eurepus-RL and https://github.com/ntnu-arl/Eurepus-design

Student Paper: yes

Submission Number: 354