Robot Dog Guided by Lab-Grown Neurons Finds Its Way to a Target on a Map

Scientists at Cortical Labs in Australia have fitted a robotic dog with lab-grown neurons that steer it toward a target on a simple map, according to ScienceAlert. The outlet described the cells as human brain cells in its headline.

The work is an early demonstration of a technology that aims to use living neurons as the basis of computer chips. The research on navigation has not yet been peer-reviewed, but it was presented at the NeurIPS 2026 conference and is available on arXiv.

How does the robot dog work?

The team grew the neurons directly on a chip fitted with electrodes that send electrical signals in both directions. That setup lets researchers translate activity in the small group of neurons into movement for the attached robot.

Brett Kagan, co-founder and CEO of Cortical Labs, called the dog a simple demonstration that helps make the opportunity tangible. He said the longer-term goal is to support robots and autonomous systems that must make decisions in unpredictable, real-world environments.

What did the Honda collaboration add?

Cortical Labs also ran a similar demonstration with Honda Research Institute Europe. In that test, everything ran in a virtual system, but the aim was the same: direct an agent around a basic grid in search of a reward.

The method was also the same. Electrical signals stimulated the neurons on a chip, and the chip sent electrical messages back. By changing the characteristics of the signal and giving feedback on the agent’s moves, the researchers trained the living chip and interpreted its movement. Over more than 4,000 hours of interactions, the team tested almost 1,300 different signal combinations.

Daniel Tanneberg, a senior scientist at Honda Research Institute Europe, said the collaboration lets researchers investigate fundamentally different approaches to intelligent agents. He said living cells as the core of computation bring many challenges but also potential in energy efficiency, robustness and adaptation.

Why does it matter?

ScienceAlert said living neurons promise advantages over conventional systems in energy efficiency and adaptability. Even though AI is sophisticated, the human brain needs far less power to run and is still much better at handling unfamiliar situations.

Kagan said robots are becoming remarkably capable at physical control, but still struggle to learn quickly or adapt when conditions change. In the longer term, ScienceAlert said, neuron-controlled robots could work in remote or hazardous conditions, or assist people with personalized responses that learn over time. Those uses remain a possibility, not a product.

Cortical Labs has previously shown clusters of neurons completing computer tasks, including playing the games Pong and Doom. The robotic dog extends that idea into the physical world.

How far along is the technology?

It is still early. ScienceAlert said the technology is not particularly fast or efficient yet, and the Honda experiment involved a fair amount of trial and error.

The next steps are to scale up the chips so they can handle more complex tasks in a shorter time, since robots in the real world will have to do far more than move around a simple grid. The researchers also want to learn more about how the neuron-based chips retain information and what kind of memory they might have.

Johnson Zhou, a research software engineer at Cortical Labs, said the challenge is not simply growing neurons on silicon. In his words, it is finding reliable ways to communicate with biological systems, train them and understand which approaches give consistent results. He said the new research provides a framework for testing those questions at scale.

Around Prague, based on reporting by ScienceAlert