Could the future of computing be biological? As we look beyond traditional silicon chips for more adaptive and energy-efficient solutions, a fascinating question arises: Can we use Synthetic Biological Intelligence (SBI)—actual living neural networks—to control physical and virtual machinery?

A pioneering new study, presented at ACM, explores this frontier by training biological neurons to solve a classic engineering challenge.

The paper, titled “Toward Controlling a Cyber-Physical System using Synthetic Biological Intelligence”, was co-authored by Pit Hofmann, Alon Loeffler, Candice Desouza, Ruifeng Zheng, Pengjie Zhou, Juan A. Cabrera, Frank H. P. Fitzek, and Brett J. Kagan. This early-stage research is a collaborative breakthrough combining the expertise of TU Dresden (including the ComNets chair), the Cluster of Excellence CeTI, and the deep-tech pioneering company Cortical Labs.

The Challenge: Teaching Living Cells to Control Systems

In engineering, the “inverted pendulum” (balancing a pole upright on a moving cart) is a classic control problem used to test how quickly and accurately a system can adapt to real-time changes. Typically, this is solved using computer algorithms.

This research, however, explores a completely different path: using a closed-loop system where living in-vitro neural networks (neurons grown on a microelectrode array) are fed sensory data and trained to send back electrical control signals to keep the virtual pendulum balanced.

Research Highlights: The Power of Biological Computing

By establishing a real-time feedback loop between digital environments and living cells, the research team highlighted several key potentials:

  • Closed-Loop Interaction: The study successfully demonstrates how electrical stimulation can “encode” the state of the pendulum for the biological network, while the neural activity is decoded in real time into mechanical actions.

  • Extreme Energy Efficiency: Biological brains process complex information using a fraction of the energy required by modern supercomputers and AI hardware. Harnessing SBI could pave the way for highly sustainable, low-power computing systems.

  • Inherent Adaptability: Unlike rigid silicon-based algorithms, living neural networks naturally reorganize, learn, and adapt to changing conditions—making them uniquely suited for unpredictable, dynamic environments.

The CeTI Impact

As CeTI continues to pioneer the future of the Tactile Internet with “Human-in-the-Loop” systems, integrating biological intelligence directly into technological control loops opens up entirely new horizons. This study is an exciting first step toward organic-synthetic hybrid systems, showing how biological neural networks can be actively integrated into cyber-physical workflows to make future technology more adaptive, resilient, and energy-efficient.

The study is available in the ACM Digital Library. Dive into the detailed methodology of this biological control system here.

Authors: Pit Hofmann, Alon Loeffler, Candice Desouza, Ruifeng Zheng, Pengjie Zhou, Juan A. Cabrera, Frank H. P. Fitzek, and Brett J. Kagan

This work is a collaboration between TU Dresden, the Cluster of Excellence CeTI, and Cortical Labs.