ModSkill: Physical Character Skill Modularization

TL;DR

ModSkill introduces a modular skill learning framework for simulated characters, improving generalization and scalability in complex motion tracking by decoupling skills into body part-specific modules and using generative adaptive sampling.

Contribution

The paper presents a novel modular skill learning approach with an attention-based skill modularization layer and active learning via generative sampling, enhancing motion tracking and skill reuse.

Findings

Outperforms existing methods in motion tracking accuracy

Enables reusable skill embeddings for various tasks

Improves generalization to larger motion datasets

Abstract

Human motion is highly diverse and dynamic, posing challenges for imitation learning algorithms that aim to generalize motor skills for controlling simulated characters. Previous methods typically rely on a universal full-body controller for tracking reference motion (tracking-based model) or a unified full-body skill embedding space (skill embedding). However, these approaches often struggle to generalize and scale to larger motion datasets. In this work, we introduce a novel skill learning framework, ModSkill, that decouples complex full-body skills into compositional, modular skills for independent body parts. Our framework features a skill modularization attention layer that processes policy observations into modular skill embeddings that guide low-level controllers for each body part. We also propose an Active Skill Learning approach with Generative Adaptive Sampling, using large motion generation models to adaptively enhance policy learning in challenging tracking scenarios. Our results show that this modularized skill learning framework, enhanced by generative sampling, outperforms existing methods in precise full-body motion tracking and enables reusable skill embeddings for diverse goal-driven tasks.