Fujitsu Develops Advanced Technology to Enhance Human–Robot Collaboration
Kawasaki, Japan – December 2, 2025 — Fujitsu today announced the development of an innovative technology designed to facilitate safer, more efficient, and seamless collaboration between humans and robots. This breakthrough, termed the spatial world model technology, enables artificial intelligence (AI) to predict future behaviors and states of various actors—such as humans, robots, and objects—within a shared physical space. The technology also supports optimized coordination among multiple robots, marking a significant advancement in the field of physical AI.
What is the Spatial World Model Technology?
The spatial world model leverages 3D scene graphs to construct a dynamic representation of a physical environment. These graphs organize all objects and actors within the space into a hierarchical structure, allowing AI to accurately interpret spatial relationships and changes over time. Unlike traditional approaches that rely on pixel-level data integration prone to errors due to varying camera perspectives and distortions, Fujitsu’s method minimizes these issues. This approach enables real-time understanding of complex and continuously changing real-world environments—a crucial capability in facilitating fluid human-robot interaction.
Predicting Future States and Behaviors
A core challenge in human-robot collaboration lies in the robot’s ability to understand human intentions and anticipate their actions to ensure smooth interactions. While earlier world model technologies could only predict immediate, localized environmental changes, Fujitsu’s new solution interprets causal relationships among humans, robots, and objects across an entire space. This holistic insight allows the system to forecast future behaviors more accurately, helping to prevent collisions and generate effective cooperative plans for multiple autonomous robots working simultaneously.
In rigorous academic benchmark tests using the JRDB-Social dataset—a recognized benchmark for estimating human behaviors from camera footage—Fujitsu’s spatial world model demonstrated a threefold increase in accuracy when estimating behavioral intentions.
Advancing Physical AI for Real-World Applications
AI has predominantly advanced through digital environments, but the emerging field of physical AI focuses on enabling AI systems to comprehend and autonomously operate within physical worlds by following physical laws. This technology is poised to address critical real-world challenges such as autonomous driving, smart factory operations, and labor shortages, particularly in Japan.
However, existing physical AI solutions largely function well only in structured settings like manufacturing plants or warehouses, where the environment and human movements are predictable. In contrast, dynamic and less structured spaces like homes or offices pose significant challenges for AI due to unpredictable human activity and frequently changing object placement. Fujitsu’s spatial world model technology aims to overcome these hurdles, paving the way for broader adoption of AI-driven human-robot collaboration in diverse environments.
Fujitsu’s Research and Future Plans
This development builds upon Fujitsu’s extensive expertise in Computer Vision, including technologies employed for analyzing human traffic flow in commercial facilities and detecting abnormal behavior for security purposes. It also integrates innovations from Fujitsu’s digital AI initiatives, such as the Kozuchi AI Agent, which collaborates autonomously with human partners.
The advancement comes as part of research activities conducted by Fujitsu’s Spatial Robotics Research Center, established in April 2025 to accelerate progress toward a future society where humans and robots coexist and collaborate harmoniously.
The spatial world model technology will be exhibited at CES 2026 in Las Vegas from January 6 to 9, 2026. Additionally, Fujitsu plans to host technical demonstrations at its headquarters throughout fiscal year 2026. ### Commitment to Sustainability and Innovation
Fujitsu’s dedication to this pioneering technology aligns with its corporate purpose “to make the world more sustainable by building trust in society through innovation.” The company also supports the United Nations’ Sustainable Development Goals (SDGs), seeking to harness technology for tangible social and economic improvements by 2030. By enhancing human-robot collaboration, Fujitsu aims to contribute to improved productivity and address workforce challenges in a variety of industries.
About Fujitsu
Fujitsu Limited is a leading digital services company headquartered in Japan, with consolidated revenues of 3.6 trillion yen (approximately US$23 billion) for the fiscal year ending March 31, 2025. Employing around 113,000 people worldwide, Fujitsu offers a wide range of services and solutions anchored in five key technologies: AI, Computing, Networks, Data & Security, and Converging Technologies. The company continuously strives to tackle some of the world’s most pressing challenges by driving digital transformation globally.
For more information, visit Fujitsu Global.
Press Contact:
Fujitsu Limited
Public and Investor Relations Division
[Contact information as per Fujitsu’s official site]
Note: All company and product names are trademarks or registered trademarks of their respective owners. Information is accurate at the time of publication and subject to change without notice.
