MirrorDuo: Reflection-Consistent Visuomotor Learning from Mirrored Demonstration Pairs

2026-06-18

AI Summary

This work addresses the limited cross-workspace generalization in behavioral cloning caused by scarce and insufficiently diverse demonstration data. To overcome this challenge, the authors propose a mirror-pair-based data augmentation and learning framework that constructs each original demonstration alongside its mirrored counterpart, effectively achieving “one collected, two obtained” data efficiency. A reflection-equivariant policy network is introduced as a structural prior to exploit this symmetry. The approach expands the training set without requiring additional real-world data, is compatible with mainstream paradigms such as behavioral cloning and diffusion policies, and enables zero-shot or few-shot (0–5 examples) skill transfer to mirrored workspaces. Experiments demonstrate that, under identical data budgets, the method significantly improves task success rates and enhances model generalization.

Abstract

Image-based behaviour cloning leverages demonstrations captured from ubiquitous RGB cameras. However, it remains constrained by the cost of collecting diverse demos, especially for generalizing across workspace variations. We propose MirrorDuo, a reflection-based formulation that operates on image, proprioception, and full 6-DoF end-effector action tuples, generating a mirrored counterpart for each original demonstration, effectively achieving "collect one, get one for free". It can be applied as a data augmentation strategy for existing learning pipelines, such as standard behaviour cloning or diffusion policy, or as a structural prior for reflection-equivariant policy networks. By leveraging the overlap between the original and mirrored domains, MirrorDuo achieves significantly improved performance under the same data budget when demonstrations are evenly distributed across both sides of the workspace. When demonstrations are confined to one side, MirrorDuo enables efficient skill transfer to the mirrored workspace with as few as zero or five demos in the target arrangement.

Problem

• behavior cloning
• demonstration diversity
• workspace generalization
• visuomotor learning
• data efficiency

Innovation

• reflection-equivariant
• visuomotor learning
• behavior cloning
• data augmentation
• mirrored demonstrations