Fujitsu’s new spatial world model technology designed to improve human–robot collaboration through real-time prediction and spatial understanding.

Himasha Dissanayake, JadeTimes Staff

H. Dissanayake is a Jadetimes news reporter covering Technology

---

Image Source: Fujitsu

Fujitsu has announced a breakthrough technology aimed at making collaboration between humans and robots safer, smoother, and significantly more efficient. The newly developed spatial world model technology enables artificial intelligence to predict future human and robot behaviors, understand dynamic spatial changes, and coordinate actions in real time. The innovation supports Fujitsu’s wider push into the fast-evolving field of physical AI.

The company will showcase the technology at CES 2026, taking place in Las Vegas from January 6–9, 2026. Additional technical demonstrations will be held at Fujitsu’s headquarters during fiscal year 2026.

Advanced 3D Scene Graphs for Real-Time Spatial Understanding

Traditional camera-based spatial analysis is often hindered by differences in camera positions, distortion, and field-of-view limitations. Fujitsu’s spatial world model overcomes these issues by using 3D scene graphs, which organize all people, robots, and objects within a space as structured graph data rather than pixel-level imagery. This allows for accurate and continuous spatial awareness—even in complex, fast-changing environments.

Figure: Illustration of the spatial world model technology.

Image Source: Fujitsu

Predicting Human Intentions and Robot Behaviors

To enable seamless collaboration, robots must anticipate human actions and interpret intention. Fujitsu’s technology models causal relationships between actors and objects, allowing AI systems to forecast future movements, avoid collisions, and generate optimal cooperative plans across multiple robots. Benchmark testing showed a threefold improvement in estimating behavioral intentions compared to existing methods.

Responding to Real-World Challenges Through Physical AI

Physical AI is increasingly crucial in industries such as logistics, smart factories, and autonomous systems. However, current solutions struggle in unstructured settings like homes, offices, and public areas where human behavior is unpredictable. Fujitsu’s approach addresses this gap and supports broader societal goals, including Japan’s labor shortages and global productivity challenges.

The innovation builds on Fujitsu’s expertise in computer vision and digital AI, including the Fujitsu Kozuchi AI Agent. It also forms part of the research at the company’s Spatial Robotics Research Center, established in April 2025 to advance human-robot coexistence technologies.

Aligned with its mission to contribute to the UN Sustainable Development Goals (SDGs), Fujitsu aims to drive forward a future in which intelligent systems enhance safety, productivity, and overall quality of life.