AXIS: A Growable Community-Driven Data Engine for Scalable Robot Manipulation
AXIS is a scalable community-driven data engine and benchmark for robot manipulation that enables browser-based teleoperation, automatically generates and validates new manipulation tasks, and transforms community-collected demonstrations into training-ready data. The AXIS dataset contains 207 diverse tasks and 50K+ trajectories. Continual pretraining on AXIS improves π0.5 by 5.8% overall, outperforms RoboCasa365 by 37.3%, and exhibits consistent scaling with increasing data volume.
Paper · arXiv:2607.21588Background
Learning effective robot manipulation policies requires diverse, high-quality demonstrations, yet existing data pipelines are difficult to scale because they rely on specialized hardware, centralized operators, or fixed task suites. The field lacks a growable, community-driven infrastructure that can continuously expand both task diversity and demonstration volume while maintaining data quality.
Core Innovation
AXIS presents a complete data engine with three key capabilities: (1) browser-based teleoperation enabling large-scale demonstration collection without specialized hardware; (2) automatic generation and validation of new manipulation tasks, allowing the task suite to grow organically; and (3) a data processing pipeline with automated success checking, quality filtering, trajectory smoothing, and visual and physics-based augmentation that transforms raw demonstrations into training-ready data. The dataset currently contains 207 diverse tasks and 50K+ trajectories. AXIS also organizes data into task snapshots and evaluates policies with a systematic held-out protocol.
Key Results
Continual pretraining on AXIS substantially improves the overall success rate of π0.5 by 5.8% and outperforms the model pretrained on RoboCasa365 by 37.3%. The approach exhibits consistent scaling with increasing data volume, with the largest gains observed under layout, sensor-noise, and camera perturbations. These results demonstrate that the quality and diversity of AXIS data translate directly into policy robustness.
Limitations
The current dataset size (207 tasks, 50K trajectories), while substantial, is still modest compared to the scale of natural language or image datasets. The browser-based teleoperation interface, while accessible, may not capture the full fidelity of kinesthetic teaching or direct robot teleoperation. Long-term community maintenance and quality control remain open challenges for any crowd-sourced platform.
Industry Implications
AXIS addresses a critical bottleneck in scalable robot learning: the availability of diverse, high-quality manipulation data. The browser-based teleoperation approach dramatically lowers the barrier to contribution, potentially accelerating data collection across the robotics community. The demonstrated improvement of foundation models like π0.5 through AXIS data suggests that community-driven data engines could become essential infrastructure for the next generation of generalist robot policies. Companies building VLA models may find AXIS a valuable complement or alternative to proprietary data pipelines.