FastLabel Inc. (Headquarters: Shinjuku-Ku, Tokyo; President and CEO: Takeshi Suzuki; hereinafter “FastLabel”), a provider of data infrastructure for robot foundation models and domain-specific Vision-Language-Action (VLA) model development, today announced the open-source release of OpenLUTRA, an application designed to streamline the collection of large-scale, high-quality datasets for AI robotics development. Background Guided by a data-centric approach to AI development, FastLabel has supported AI robotics development for many organizations, including major Japanese companies and research institutions, across every stage of the AI development process, from data collection and data processing to model development and evaluation. Through the execution of numerous projects, FastLabel has identified the following challenges in the field of data collection: The inclusion of data that does not meet the required criteria—such as cases where the robot's initial pose differs from the specified conditions or where tasks are not completed within the standard execution time—cannot be detected, resulting in time spent on data collection being wasted. Delays or missing camera data and sensor data are discovered only after data collection has been completed, rendering the collected data unusable. Data recording and quality control mechanisms are difficult for non-engineer operators to use, requiring engineers to provide on-site support and making it difficult to scale data collection. To address the challenges described above and provide strong support for AI robotics research, development, and social implementation from the point where data is generated, FastLabel has open-sourced OpenLUTRA, an application that enables high-quality, scalable data collection. OpenLUTRA Interface Key Features of OpenLUTRA ■Real-Time Anomaly Monitoring By monitoring data delays, missing data, and system health metrics in real time, OpenLUTRA enables early recovery when anomalies occur. ■Automated Quality Assessment (Curation Function) OpenLUTRA automatically analyzes stored data, evaluates whether it meets predefined quality criteria, and assigns tags accordingly. The quality assessment logic is fully customizable, allowing users to implement custom evaluation logic tailored to the requirements of individual use cases. ■Intuitive UI/UX OpenLUTRA features a user interface that can be easily operated by non-engineer robot operators, providing an intuitive user experience that enables smooth operation in the field. ■Support for a Wide Range of Robots and Devices OpenLUTRA is compatible with ROS 2, a standard robotics development environment. With minimal setup requirements, the application can be deployed quickly across diverse robotics development and data collection environments. About OpenLUTRA GitHub: https://github.com/fastlabel/open-lutra License: Apache License, Version 2.0 Version: The current release is a beta version. An official release is scheduled within the next few months. This application is available for commercial use, and all rights to the data obtained through its use belong to the user. Notes *1 Operator An individual who operates a robot using a controller. *2 Curation: Curation refers to the process of selecting and organizing datasets based on specific objectives or predefined rules. In the context of AI development, it is used for purposes such as removing low-quality or duplicate data, extracting data that meets specific criteria, and adjusting data distribution. Join Us as a Physical AI Engineer At FastLabel, we are actively recruiting professional teammates who sincerely approach technology and customers, and who want to challenge solving real-world social issues through the social implementation of AI robotics. Main Roles: Investigation and verification of cutting-edge robot foundation models, learning strategies, and data collection methodologies Training and evaluation of robot foundation models, and realization of robot control using these models Design and systematization of cutting-edge data collection methodologies Resolution of customer issues through implementation in real-world environments Career Opportunities: https://herp.careers/v1/fastlabel/dHh35DUK1nXO Engineering Blog: https://fastlabel.hatenablog.com/ About FastLabel FastLabel, Inc. is committed to building data infrastructure that enables data-centric AI development. The company provides end-to-end support across the AI development lifecycle, including data collection and generation, annotation, model development, and DataOps implementation. In recent years, FastLabel has expanded its focus to the field of physical AI, including robotics, and has been developing data pipelines that support the creation of robot foundation models and Vision-Language-Action (VLA) models. Company Overview Company Name: FastLabel Inc. Headquarters: Shinjuku Sumitomo Building 24F, 2-6-1 Nishi-Shinjuku, Shinjuku-ku, Tokyo, Japan Representative: Takeshi Suzuki, President and CEO Business: Provider of professional services and products supporting Data-centric AI development Website: https://fastlabel.ai/service/robotics