SkillVLA: Tackling Combinatorial Diversity in Dual-Arm Manipulation via Skill Reuse

TL;DR

SkillVLA introduces a framework for dual-arm manipulation that enables skill reuse, significantly improving the ability to recombine learned skills for complex tasks and outperforming existing models in success rate.

Contribution

The paper presents SkillVLA, a novel framework that explicitly supports skill reuse in dual-arm manipulation, addressing the challenge of combinatorial diversity in vision-language-action models.

Findings

Success rate increased from 0% to 51%.

Improved skill composition for complex tasks.

Strong performance on cooperative and long-horizon tasks.

Abstract

Recent progress in vision-language-action (VLA) models has demonstrated strong potential for dual-arm manipulation, enabling complex behaviors and generalization to unseen environments. However, mainstream bimanual VLA formulations largely overlook the critical challenge of combinatorial diversity. Different pairings of single-arm behaviors can induce qualitatively distinct task behaviors, yet existing models do not explicitly account for this structure. We argue that effective bimanual VLAs should support skill reuse - the ability to recombine previously learned single-arm skills across novel left-right pairings - thereby avoiding the need to separately learn every possible combination. Current VLA designs entangle skills across arms, preventing such recomposition and limiting scalability. To address this limitation, we propose SkillVLA, a framework explicitly designed to enable skill reuse in dual-arm manipulation. Extensive experiments demonstrate that SkillVLA substantially improves skill composition, increasing overall success rate from 0% to 51%, and achieves strong performance on cooperative and long-horizon tasks.