Advancing physical AI: Building robots that learn from failure

I enjoy sports and have played tennis since the fifth grade through university. While I didn’t have a clear career path in mind, I hadn’t been able to achieve my goal of competing in a national tennis tournament in high school, so I joined the School of Informatics in university seeking an environment where I could continue to compete while balancing academics. My interest in AI began to grow through my undergraduate research. I wanted to combine AI, which was gaining increasing attention at the time, with my longtime experience in tennis, going on to pursue research under the theme of “AI for Tennis”. In tennis, players make strategic decisions based on their opponent’s position and the likely direction of the next shot. In my thesis, I used AI to explore whether this information could be quantified through motion capture and applied to strategic decision-making. Specifically, I collected and analyzed data such as players’ positions (where they stand and where they hit the ball), as well as the ball’s trajectory and bounce points, aiming to contribute to performance improvement in matches. Though I continued to struggle to choose a career even after entering graduate school, my research allowed me to discover the excitement of visualizing phenomena previously understood only intuitively. I also became fascinated by the potential of technology to expand human capabilities, which inspired me to contribute to the creation of new technologies as a researcher.

I am currently a part of the Physical AI laboratory, which was established in April 2026. The laboratory aims to realize a world where humans can safely coexist and collaborate with a wide range of robots across diverse real-world environments. It also promotes a vision of evolving robots to contribute to the development of society as a whole by incorporating on-site knowledge, advancing collaborative capabilities, and sharing these outcomes across locations and operations. To achieve this vision, we are developing Fujitsu Kozuchi Physical OS (*1), a platform that integrates robots, sensors, systems, and physical spaces to enable the robots to correctly understand tasks and execute the appropriate actions based on operational instructions.

Within this initiative, I leverage my experience in computer vision and video understanding to conduct research on robotic intelligence (*2). My current focus is on technologies that enable the reliable deployment of Vision-Language-Action (VLA) models (*3), which serve as the decision-making component of robotic systems. Key challenges include handling errors that arise during long-horizon tasks and adapting to unexpected changes in the environment. To address these challenges, I aim to develop technologies that enable robots to proactively avoid failures and, when they occur, accurately identify their causes and recover autonomously. I believe that overcoming these challenges will pave the way for technologies that can operate reliably in real-world environments.

My research on long-term video understanding, which I had been working on for some time, was accepted at CVPR 2026 (*4)—one of the most prestigious international conferences in the field. In June, I presented a poster of my work on-site. Although I had submitted many papers, I often received reviewer feedback that the key contributions were not clearly communicated and that the paper was difficult to grasp quickly. Despite achieving significant improvements in inference performance for long-term video understanding through repeated experimentation with my senior colleagues, I felt a strong sense of regret that our work was not fully recognized.

Through the process of writing these papers, I came to realize that my responsibility extended beyond developing the technology itself to communicate its value effectively. I rethought my approach by actively involving researchers from beyond the team and increasing opportunities for discussion and review. This led to substantial improvements in the clarity of the paper’s structure and arguments as the technology itself continued to be refined, ultimately leading to the paper’s acceptance. Through this experience, I was reminded that the value of technology is brought out by people, and that being able to shift one’s mindset is important to achieving goals.

In a separate project, I took on responsibilities beyond research, such as independently bringing equipment to an overseas exhibition and conducting demonstrations. These experiences were highly stimulating and provided valuable learning opportunities. Although I encountered technical issues during the demonstration, I addressed them through continuous improvements to the system. Seeing visitors engage with the technology and respond positively was highly rewarding. Through these experiences, I came to understand that technology is not only about building systems, but also about effectively communicating and delivering value. This realization has shaped how I approach research and development today.

I have two main goals for the future. The first is to continue seeking experiences that broaden my horizons and deepen my understanding of the world, both through research and beyond it. As part of this goal, I am considering pursuing a doctoral degree. By engaging with diverse perspectives and new environments through research, I hope to continue growing both professionally and personally. My second goal is to make research a more accessible and appealing career option for a wider range of people. In particular, I hope to encourage those who may feel uncertain about pursuing a path in science or research. I did not originally envision myself becoming a researcher either. However, through a series of small opportunities and words of encouragement, I discovered how exciting research could be and ultimately found my place in it. In the future, I hope to provide similar opportunities for others and help them realize paths they may not yet see for themselves.

“A well-balanced researcher”—that was my first impression when Natsuki joined our team at Fujitsu. She has strong presentation skills that allow her to explain ideas easily and clearly to people; solid communication skills, including in English; and a proactive attitude toward advancing her research through hands-on work. As a member of the research team, she has leveraged these strengths, gained wide-ranging experiences, and achieved significant growth. Now is the time for her to move beyond being well-balanced and stand out as a researcher. I hope she will continue to break through barriers and set herself apart. Here at the Physical AI Laboratory, we will provide the best possible research environment to support her in doing so.
(Genta Suzuki, Head of the Physical AI Laboratory)

Titles, numerical values, and proper nouns in this document are those reported when this interview was made.