Public Release of a Robot Motion Dataset and Robot Foundation Model from an FY2025 Project Selected under GENIAC

On July 31, 2026, the AI Robot Association (AIRoA) publicly released the following deliverables as outcomes of the project titled “Study on the Construction of Robot Motion Datasets for Training to Accelerate AI-Based General-Purpose Robot Development, and Verification of Their Potential Application to Robot Foundation Models.” The project was selected under the “Post-5G Information and Communication System Infrastructure Enhancement R&D Program / Survey on the Utilization of Data and Generative AI to Accelerate Generative AI Development (GENIAC),” jointly implemented by the Ministry of Economy, Trade and Industry (METI) and the New Energy and Industrial Technology Development Organization (NEDO).

• Data: Approximately 5,000 hours of cleansed data (68 task types)
• Model: A continually pretrained model based on the open-source VLA model (π0.5), together with its training source code

Through this release, AIRoA aims to support the research, development, and validation of robot foundation models by research institutions and development companies in both Japan and abroad, and contribute to the development of a future data ecosystem.

The deliverables released here were obtained through research, development, and validation using the HSR (Human Support Robot) mobile manipulator.

Please refer to the technical documentation on each release page for information on supported environments, usage instructions, and other details.

Released Materials and Where to Obtain Them

Robot Motion Dataset

AIRoA MoMa 5k
https://huggingface.co/datasets/airoa-org/airoa-moma-5k

Model Weights

pi05_hsr_base
https://huggingface.co/airoa-org/airoa-pi05-hsr-base

Model Training Source Code

AiroPi
https://github.com/airoa-org/airopi-share

Each deliverable is provided under the license and terms of use specified on its respective release page. Please review the applicable conditions of use and disclaimers before use.

These results were achieved with support from GENIAC (Generative AI Accelerator Challenge), a national project jointly implemented by METI and NEDO to strengthen Japan’s generative AI development capabilities and facilitate its societal implementation.

Contact Information

Inquiries via the AIRoA official website
https://www.airoa.org/contact/