JTEKT, Upstage AI, and Karakuri's Co-creation Project Selected for 'Physical AI Development Support Program by AWS Japan'

JTEKT Corporation (Headquarters: Kariya City, Aichi Prefecture; President: Yoshihito Kondo; hereinafter "JTEKT") announces that a project aiming to implement physical AI in manufacturing sites, jointly promoted with Upstage AI, Inc. (Headquarters: Minato-ku, Tokyo; Representative Director: Hiroyuki Matsushita; hereinafter "Upstage") and Karakuri Inc. (Headquarters: Chuo-ku, Tokyo; Representative Director and CEO: Shimon Oda; hereinafter "Karakuri"), has been selected for the "Physical AI Development Support Program by AWS Japan" conducted by Amazon Web Services Japan G.K. (hereinafter "AWS Japan"). Through selection in this program, the three companies will verify the potential of utilizing physical AI in the manufacturing industry and aim for the social implementation of autonomous and adaptive production systems. Based on the mission of "Connecting technologies to bring smiles to the Earth and all people who work," the JTEKT Group has set the JTEKT Group 2030 Vision as its goal for 2030: "A solution provider that creates the future of a mobility society through manufacturing and manufacturing equipment." To transform into this solution provider, we are engaging in corporate activities with a dual focus on adding high value to existing products and challenging new domains. ## 1. Background of the Project While the Japanese manufacturing industry possesses high technological capabilities, it faces structural challenges such as labor shortages due to a declining birthrate and aging population, as well as the succession of skilled techniques. To solve these issues, JTEKT aims to implement physical AI that can autonomously and adaptively respond to changing environments for robots and production equipment by utilizing the diverse and highly accurate data generated on-site and the actual production environment. ## 2. Overview of Joint Development In this project, we aim to realize the transition to an autonomous and adaptive production system in JTEKT's factories through the following approaches: 1. Utilization of Vision Language Action (VLA): By comprehensively learning and inferring images, language, and actions using multimodal generative AI, we will realize the automation and optimization of complex manufacturing processes. 2. Reduction of prototype costs through high-speed learning in virtual space (Sim2Real): By transferring the learning results in the simulation environment to the real environment, we will significantly reduce the development cost and prototyping period of AI models. 3. Capitalizing on-site data and making it AI-Ready: We will convert the diverse and highly accurate data accumulated in JTEKT's factories into a format that AI can utilize to the fullest, and construct a dedicated model based on the on-site data. ### - JTEKT - Demonstration field and