Fujitsu and Carnegie Mellon Open Physical AI Research Hub

• Fujitsu and Carnegie Mellon University launched a joint research center to advance core physical AI technologies and support deployment across industries facing productivity, labor, and safety challenges.
• Technologies developed through the center will be gradually incorporated into Fujitsu Kozuchi Physical OS starting in fiscal year 2026 to improve robot coordination and real-world adaptability.

Fujitsu and Carnegie Mellon University announced the Fujitsu-Carnegie Mellon Physical AI Research Center, a new joint hub focused on advancing physical AI. The partnership will support research across robotics, AI, simulation, and human-robot interaction to help address challenges such as improving productivity, mitigating labor shortages, and ensuring safety in real-world operations.

The center will operate through Carnegie Mellon’s 14,000-square-meter Robotics Innovation Center at Hazelwood Green in Pittsburgh, which opened in February and is designed to connect fundamental research with commercial deployment. Faculty from robotics, machine learning, language technologies, engineering, human-computer interaction, and philosophy will work alongside Fujitsu scientists and engineers on action generation, spatial perception, multi-robot coordination, human-robot collaboration, and bridging simulation with real-world environments.

Fujitsu said the research will support Fujitsu Kozuchi Physical OS, a platform that integrates robots, sensors, systems, and physical spaces. Technologies from the center are scheduled to be gradually incorporated into the platform starting in fiscal year 2026, with both organizations stating they will continue advancing physical AI research to support a society where humans and robots collaborate and to contribute to a sustainable and resilient society.

🌀 Tom’s Take:

By linking Fujitsu’s deployment goals with Carnegie Mellon’s academic depth, the center creates a stronger path for physical AI research to move beyond experiments and into real-world systems.

Source: Fujitsu