AIRoA Selected for GENIAC(*1), a National Project Jointly Implemented by METI and NEDO – Toward Building a Data Ecosystem That Continuously Expands the Collection and Use of Robot Action Data in Real-World Homes

Aiming to establish a virtuous cycle of data collection, model improvement, and the accumulation of operational insights through continuous operation in real-world homes

We have been selected as a participating organization under the “Post-5G Information and Communication System Infrastructure Enhancement R&D Program / Research and Development on the Establishment of Data Ecosystems (GENIAC),” jointly implemented by the Ministry of Economy, Trade and Industry (METI) and the New Energy and Industrial Technology Development Organization (NEDO). The selected project is titled “Research and Development of a Data Ecosystem for the Self-Sustaining Expansion of Robot Action Data in Real Homes.”

(*1) GENIAC (Generative AI Accelerator Challenge) is a national project jointly implemented by the Ministry of Economy, Trade and Industry (METI) and the New Energy and Industrial Technology Development Organization (NEDO) to strengthen Japan’s generative AI development capabilities and facilitate its societal implementation.  

METI announcement: https://www.meti.go.jp/press/2026/07/20260702001/20260702001.html

NEDO announcement: https://www.nedo.go.jp/koubo/CD3_100430.html

Background: The Importance of Data from Real-World Homes for the Real-World Deployment of Home Robots

As Japan faces a declining birthrate, an aging population, and labor shortages, expectations are growing for home robots that can assist with household chores, monitoring, and daily living support.

Home environments, however, differ from household to household in terms of floor plans, the placement of furniture and everyday items, lifestyles, and the procedures used to perform household tasks. Compared to factories and research facilities, homes are highly variable environments, making it difficult for robots to operate safely and flexibly.

Furthermore, continuously operating robots and collecting data in real-world homes also involves challenges that cannot be solved through technological development alone, such as privacy, safety, resident consent management, teleoperation, and robot maintenance. It is not easy for individual companies to develop the necessary data resources and operational insights at scale.

To date, AIRoA has been developing infrastructure for collecting and processing robot action data; developing and validating Vision-Language-Action (VLA) models; collecting data using mobile manipulators, YUBI (Yielding Universal Bidigital Interface) and other tools; and conducting demonstrations in simulated living environments, such as the one established at the National Museum of Emerging Science and Innovation (Miraikan).

Following its selection for this project, AIRoA will extend these capabilities to real-world home environments and develop shared data and operational infrastructure that can be used by multiple robotics companies and AI developers. Through these efforts, AIRoA will advance R&D on robot foundation models and promote the use of robots in real-world environments.

Project Objectives

Under this project, robots will be deployed in real-world homes to collect robot action data together with user feedback, including instructions and evaluations.

Specifically, AIRoA plans to deploy robots in a total of 35 households, including households in Toyota Woven City—which is built around concepts including A Living Laboratory™—and collect approximately 7,000 hours of robot action data.

AIRoA will operate robots on an ongoing basis by combining autonomous task execution with teleoperation, and use the collected data to train, evaluate, and improve VLA models. Enhancing model performance and robot autonomy will expand the range of environments in which robots can operate, enabling the collection of further data.

In addition, AIRoA will develop the following components as an integrated infrastructure:

• Applications for users and operators
• Infrastructure for collecting and managing data in real homes
• An operational framework covering teleoperation and robot maintenance
• Mechanisms for safety, privacy, and consent management
• Quality control and rights clearance processes for collected data

The data and knowledge obtained through this project will be developed into datasets, guidelines, and operational frameworks that can subsequently be used by robotics companies and AI developers seeking to build on these results.

AIRoA aims not only to collect data but also to establish a system in which robots can be continuously operated across diverse home environments, creating a virtuous cycle of data accumulation and model improvement.

Evaluation in the Selection Process

During NEDO’s review, the proposal was recognized for its importance and novelty in the field of home robotics, the feasibility enabled by AIRoA’s proven teleoperation system, and its integrated approach spanning the collection of useful datasets that form the basis of VLA models through to VLA development.

Next Steps

At present, the project is scheduled to run from September 2026 through August 2027. In preparation for the project launch, AIRoA will establish the project implementation structure, data collection and management infrastructure, and operational framework for deployment in real homes.