LatentMimic: Terrain-Adaptive Locomotion via Latent Space Imitation

TL;DR

LatentMimic is a novel framework for quadruped locomotion that decouples style from geometry, enabling terrain adaptation while preserving motion style, validated across multiple terrains and styles.

Contribution

It introduces a latent space imitation approach that relaxes pose-tracking constraints, allowing style preservation and terrain adaptability in quadruped locomotion.

Findings

Achieves higher terrain traversal success rates than state-of-the-art methods.

Effectively maintains gait style across diverse terrains.

Demonstrates versatility across four locomotion styles and terrains.

Abstract

Developing natural and diverse locomotion controllers for quadruped robots that can adapt to complex terrains while preserving motion style remains a significant challenge. Existing imitation-based methods face a fundamental optimization trade-off: strict adherence to motion capture (mocap) references penalizes the geometric deviations required for terrain adaptability, whereas terrain-centric policies often compromise stylistic fidelity. We introduce LatentMimic, a novel locomotion learning framework that decouples stylistic fidelity from geometric constraints. By minimizing the marginal latent divergence between the policy's state-action distribution and a learned mocap prior, our approach provides a conditional relaxation of rigid pose-tracking objectives. This formulation preserves gait topology while permitting independent end-effector adaptations for irregular terrains. We further introduce a terrain adaptation module with a dynamic replay buffer to resolve the policy's distribution shifts across different terrains. We validate our method across four locomotion styles and four terrains, demonstrating that LatentMimic enables effective terrain-adaptive locomotion, achieving higher terrain traversal success rates than state-of-the-art motion-tracking methods while maintaining high stylistic fidelity.