Exciting Action: Investigating Efficient Exploration for Learning Musculoskeletal Humanoid Locomotion

TL;DR

This paper explores how adversarial imitation learning can efficiently train musculoskeletal humanoid models to achieve natural walking and running gaits with minimal demonstrations, overcoming challenges of high-dimensional control.

Contribution

It introduces a novel application of adversarial imitation learning to musculoskeletal locomotion, providing solutions for effective exploration and gait learning with limited demonstrations.

Findings

Achieved natural-looking gaits with few demonstrations

Validated methodology on a complex simulated humanoid model

Demonstrated effective exploration in high-dimensional action spaces

Abstract

Learning a locomotion controller for a musculoskeletal system is challenging due to over-actuation and high-dimensional action space. While many reinforcement learning methods attempt to address this issue, they often struggle to learn human-like gaits because of the complexity involved in engineering an effective reward function. In this paper, we demonstrate that adversarial imitation learning can address this issue by analyzing key problems and providing solutions using both current literature and novel techniques. We validate our methodology by learning walking and running gaits on a simulated humanoid model with 16 degrees of freedom and 92 Muscle-Tendon Units, achieving natural-looking gaits with only a few demonstrations.