MOTIF: Learning Action Motifs for Few-shot Cross-Embodiment Transfer

TL;DR

MOTIF introduces a novel approach for few-shot cross-embodiment transfer in robotics by learning embodiment-agnostic action motifs, enabling efficient adaptation across different robot embodiments with minimal data.

Contribution

The paper proposes MOTIF, a method that learns shared action motifs using vector quantization and alignment techniques, facilitating effective cross-embodiment transfer with few demonstrations.

Findings

Outperforms baselines by 6.5% in simulation

Achieves 43.7% improvement in real-world transfer

Validates effectiveness in both simulation and real environments

Abstract

While vision-language-action (VLA) models have advanced generalist robotic learning, cross-embodiment transfer remains challenging due to kinematic heterogeneity and the high cost of collecting sufficient real-world demonstrations to support fine-tuning. Existing cross-embodiment policies typically rely on shared-private architectures, which suffer from limited capacity of private parameters and lack explicit adaptation mechanisms. To address these limitations, we introduce MOTIF for efficient few-shot cross-embodiment transfer that decouples embodiment-agnostic spatiotemporal patterns, termed action motifs, from heterogeneous action data. Specifically, MOTIF first learns unified motifs via vector quantization with progress-aware alignment and embodiment adversarial constraints to ensure temporal and cross-embodiment consistency. We then design a lightweight predictor that predicts these motifs from real-time inputs to guide a flow-matching policy, fusing them with robot-specific states to enable action generation on new embodiments. Evaluations across both simulation and real-world environments validate the superiority of MOTIF, which significantly outperforms strong baselines in few-shot transfer scenarios by 6.5% in simulation and 43.7% in real-world settings. Code is available at https://github.com/buduz/MOTIF.