Tree Learning: A Multi-Skill Continual Learning Framework for Humanoid Robots

TL;DR

Tree Learning is a hierarchical continual learning framework for humanoid robots that prevents catastrophic forgetting and supports multi-skill adaptation using parameter inheritance and multi-modal mechanisms.

Contribution

It introduces a root-branch hierarchical parameter inheritance mechanism and a multi-modal adaptation method for efficient multi-skill learning in humanoid robots.

Findings

Tree Learning achieves higher rewards than simultaneous multi-task training.

It maintains a 100% skill retention rate across various skills.

The framework enables seamless multi-skill switching and real-time control.

Abstract

As reinforcement learning for humanoid robots evolves from single-task to multi-skill paradigms, efficiently expanding new skills while avoiding catastrophic forgetting has become a key challenge in embodied intelligence. Existing approaches either rely on complex topology adjustments in Mixture-of-Experts (MoE) models or require training extremely large-scale models, making lightweight deployment difficult. To address this, we propose Tree Learning, a multi-skill continual learning framework for humanoid robots. The framework adopts a root-branch hierarchical parameter inheritance mechanism, providing motion priors for branch skills through parameter reuse to fundamentally prevent catastrophic forgetting. A multi-modal feedforward adaptation mechanism combining phase modulation and interpolation is designed to support both periodic and aperiodic motions. A task-level reward shaping strategy is also proposed to accelerate skill convergence. Unity-based simulation experiments show that, in contrast to simultaneous multi-task training, Tree Learning achieves higher rewards across various representative locomotion skills while maintaining a 100% skill retention rate, enabling seamless multi-skill switching and real-time interactive control. We further validate the performance and generalization capability of Tree Learning on two distinct Unity-simulated tasks: a Super Mario-inspired interactive scenario and autonomous navigation in a classical Chinese garden environment.