Switch: Learning Agile Skills Switching for Humanoid Robots

arXiv cs.RO / 4/17/2026

📰 NewsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper proposes Switch, a hierarchical multi-skill framework for humanoid robots to transition between locomotion skills smoothly at any moment.
Switch uses a Skill Graph built from kinematic similarity in multi-skill motion data to define feasible cross-skill transitions.
A whole-body tracking policy is trained with deep reinforcement learning over the skill graph, enabling stable execution of diverse skills.
An online skill scheduler performs real-time graph search when switching skills or when tracking deviates, selecting an optimal feasible transition path for safety and responsiveness.
Experiments show high success rates for agile transitions while preserving strong motion imitation performance.

Abstract

Recent advancements in whole-body control through deep reinforcement learning have enabled humanoid robots to achieve remarkable progress in real-world chal lenging locomotion skills. However, existing approaches often struggle with flexible transitions between distinct skills, cre ating safety concerns and practical limitations. To address this challenge, we introduce a hierarchical multi-skill system, Switch, enabling seamless skill transitions at any moment. Our approach comprises three key components: (1) a Skill Graph (SG) that establishes potential cross-skill transitions based on kinematic similarity within multi-skill motion data, (2) a whole-body tracking policy trained on this skill graph through deep reinforcement learning, and (3) an online skill scheduler to drive the tracking policy for robust skill execution and smooth transitions. For skill switching or significant tracking deviations, the scheduler performs online graph search to find the optimal feasible path, which ensures efficient, stable, and real-time execution of diverse locomotion skills. Comprehensive experiments demonstrate that Switch empowers humanoid to execute agile skill transitions with high success rates while maintaining strong motion imitation performance.