Dexora: Open-source VLA for High-DoF Bimanual Dexterity

TL;DR

Dexora is an open-source vision-language-action system enabling high-DoF dual-arm, dual-hand manipulation with a hybrid teleoperation pipeline, large datasets, and a novel training approach, advancing embodied AI capabilities.

Contribution

Introducing Dexora, the first open-source VLA system for high-DoF bimanual manipulation, with a hybrid teleoperation pipeline, large datasets, and a data-quality-aware training method.

Findings

Dexora outperforms baselines on dexterous benchmarks with 66.7% success.

Achieves 90% success on basic manipulation tasks.

Demonstrates robust generalization across different embodiments.

Abstract

Vision-Language-Action (VLA) models have recently become a central direction in embodied AI, but current systems are restricted to either dual-gripper control or single-arm dexterous hand manipulation. While low-dimensional gripper control can often be handled with simpler methods, high-dimensional dexterous hand control benefits greatly from full end-to-end VLA learning. In this work, we introduce Dexora, the first open-source VLA system that natively targets dual-arm, dual-hand high-DoF manipulation. We design a hybrid teleoperation pipeline that decouples gross arm kinematics (captured with a custom exoskeleton backpack) from fine finger motion (markerless hand tracking via Apple Vision Pro), and that drives both a physical dual-arm dual-hand platform and an identical MuJoCo digital twin. Using that interface, we assemble a large training corpus: an embodiment-matched synthetic corpus (100K simulated trajectories, 6.5M frames) and a real-world dataset of 10K teleoperated episodes (2.92M frames). To mitigate noisy teleoperation demonstrations, we propose a data-quality-aware training recipe: an offline discriminator provides clip-level weights for diffusion-transformer policy training, down-weighting low-quality demonstrations. Empirically, Dexora outperforms competitive VLA baselines on both basic and dexterous benchmarks (e.g., average dexterous success 66.7% vs. 51.7%), attains 90% success on basic tasks, and shows robust out-of-distribution and cross-embodiment generalization. Ablations confirm the importance of real data and the discriminator for dexterity.