Young Researcher: Associate Professor Kohei NAKAJIMA

Associate Professor　Department of Mechano-Informatics, Graduate School of Information Science and Technology / Department of Mechano-Informatics｜Kuniyoshi-Nakajima Lab (Laboratory for Intelligent Systems and Informatics)

Biography

I conduct research on soft robotics. Soft robots are robots with soft, flexible bodies inspired by those of living organisms. For example, the robot arm shown in Fig. 1a is a silicone robotic arm inspired by an octopus.
Living organisms have soft bodies and are able to adaptively control their behavior in response to changes in their environment. In contrast, controlling the behavior of robots in dynamic, real-world environments remains highly challenging. So, how can we equip robots with the kind of flexible adaptability found in living organisms?
I focused on the diverse dynamics generated by the bodies of soft robots. I showed that the dynamics of a soft body can, in themselves, perform information processing [1] (Fig. 1b). For example, by moving the octopus arm shown in Fig. 1a, we can simultaneously perform Boolean logic operations. This is enabled by a technique called physical reservoir computing [2].
Based on this perspective, I have developed a series of soft robots and demonstrated that their softness can be exploited not only for physical movement but also for sensing and control (e.g., [3]).

[1] Nakajima, K., Li, T., Hauser, H., & Pfeifer, R. (2014). Exploiting short-term memory in soft body dynamics as a computational resource. Journal of The Royal Society Interface, 11(100), 20140437.
[2] Nakajima, K. (2020). Physical reservoir computing—an introductory perspective. Japanese Journal of Applied Physics, 59(6), 060501.
[3] Akashi, N., Kuniyoshi, Y., Jo, T., Nishida, M., Sakurai, R., Wakao, Y., & Nakajima, K. (2024). Embedding bifurcations into pneumatic artificial muscle. Advanced Science, 11(25), 2304402.

Fig. 1

More recently, I have expanded this approach beyond soft robotics by developing physical reservoirs that use living organisms themselves as computational resources, including a learning machine based on Tetrahymena [1] and a jellyfish cyborg [2].
I have also applied this idea to a variety of physical systems and conducted theoretical studies to investigate what kinds of dynamics give rise to what kinds of computational capabilities. Complex dynamical systems can also be found in artificial systems. For example, I demonstrated that highly complex dynamics known as transient chaos emerge within large language models [3], and have conducted research comparing these dynamics with those of the human brain [4].

[1] Ushio, M., Watanabe, K., Fukuda, Y., Tokudome, Y., & Nakajima, K. (2023). Computational capability of ecological dynamics. Royal Society open science, 10(4), 221614.
[2] Owaki, D., Austin, M., Ikeda, S., Okuizumi, K., & Nakajima, K. (2025). Harnessing natural embodied intelligence for spontaneous jellyfish cyborgs. Nature Communications, 16(1), 4642.
[3] Inoue, K., Ohara, S., Kuniyoshi, Y., & Nakajima, K. (2022). Transient chaos in bidirectional encoder representations from transformers. Physical Review Research, 4(1), 013204.
[4] Watanabe, T., Inoue, K., Kuniyoshi, Y., Nakajima, K., & Aihara, K. (2025). Comparison of large language model with aphasia. Advanced Science, 12(22), 2414016.

My hobby is talking with people. Please tell me about the things you are interested in. I look forward to discussing them with all of you!

Kuniyoshi-Nakajima Lab (Laboratory for Intelligent Systems and Informatics): https://www.isi.imi.i.u-tokyo.ac.jp/
Physical Intelligence Lab: https://www.kohei-nakajima.com/

 *Affiliations and titles are as of the time of the interview.