Learning Context-Aware Motion Priors for Humanoid Control

2026-08-04

AI Summary

Traditional motion priors struggle to dynamically select context-relevant reference motions based on task-specific cues, often introducing irrelevant or conflicting guidance. This work proposes a Context-Aware Motion Prior (CMP) framework that learns the compatibility between task context and reference motions through policy advantage signals, and employs demonstration-supervised dynamic loss reweighting to train a lightweight context-conditioned adapter. The approach achieves adaptive alignment between motion priors and task context without requiring manual skill labels, dataset partitioning, or a predefined skill discovery phase. Experiments demonstrate that CMP significantly improves performance and sample efficiency across five humanoid control tasks, learns interpretable context-motion associations, and exhibits robustness to imbalanced reference motion distributions.

Abstract

Motion priors provide powerful guidance for learning naturalistic humanoid behaviors. However, existing methods typically learn a general, task-agnostic prior from the entire reference dataset and apply it uniformly throughout policy training. As a result, the prior cannot distinguish which reference motions are relevant to the current task context, potentially providing irrelevant or conflicting guidance. We present Context-Aware Motion Priors (CMP), a framework that adapts a general motion prior to the current task context without manual skill labels, dataset partitioning, or a separate skill discovery stage. Specifically, CMP learns context-motion compatibility using high-advantage policy rollouts, while a demonstration-based objective keeps the learned relevance grounded in the reference distribution. The resulting relevance scores reweight reference supervision for training a lightweight context-conditioned adapter. To evaluate the effectiveness and generality of CMP, we instantiate it with both Adversarial Motion Priors and Score-Matching Motion Priors. Across five humanoid control tasks, CMP consistently improves task performance and sample efficiency, learns meaningful context-motion alignment, and remains robust to imbalanced reference distributions. These results show that adapting motion priors to task contexts provides more relevant guidance for humanoid policy learning.

Problem

• motion priors
• humanoid control
• task context
• reference motions
• context-awareness

Innovation

• Context-Aware Motion Priors
• Humanoid Control
• Motion Prior Adaptation
• Policy Learning
• Demonstration-based Learning