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

2026-04-14

AI Summary

This work addresses catastrophic forgetting in continual learning of diverse motor skills for humanoid robots by proposing a lightweight, tree-structured hierarchical parameter inheritance architecture. The root node encodes a general motion prior, and knowledge is propagated to skill-specific branches through parameter reuse, fundamentally mitigating forgetting. A multimodal feedforward mechanism integrating phase modulation and motion interpolation unifies the handling of both periodic and aperiodic actions. Task-level reward shaping is further incorporated to accelerate policy convergence. Evaluated in Unity simulations, the approach achieves 100% retention across multiple locomotion skills with higher cumulative rewards, enables seamless skill switching and real-time interaction, and demonstrates strong generalization in challenging environments such as Mario-style platforming and Chinese garden navigation tasks.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Machine Learning: Life-Long and Continual Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

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.

Problem

• continual learning
• catastrophic forgetting
• multi-skill learning
• humanoid robots
• embodied intelligence

Innovation

• Tree Learning
• continual learning
• parameter inheritance
• multi-modal adaptation
• reward shaping