Researchers at NTNU are developing a robot that will be controlled by living brain cells. A cyborg (cybernetic organism) is a combination of machine and living tissue.
Cyborgs have been part of our fantasy world since the start of the science fiction genre. The interface between living organisms and machines has been used to frighten and entertain, as well as to explore the boundary between man and technology.
A cyborg is about to become reality at NTNU. Scientists and students are working on constructing an interactive social robot that will connect to a biological “brain.”
The cyborg may become a familiar figure on campus down the road – while also helping researchers understand how to repair brain damage and construct new types of computers.
It probably won’t be taking over the world anytime soon. But it will become a cyborg. '
A cyborg that wants to be your friend
“We start with the machine, and then we ‘bring it to life’ by adding biological neurons. Then we have to keep them alive and get them to communicate with a computer,” says researcher Stefano Nichele at NTNU’s Department of Computer Science.
He coordinates NTNU Cyborg, a project that includes NTNU research groups in various disciplines – ranging from computer technology and cybernetics to neuroscience, ethics and design.
The goal is to build a social robot that will stroll around campus and interact with students and staff. It will be able to approach and greet people, recognize them, and even send them a friend request on Facebook. Combining the robot and the living neural network makes this structure unique. According to Nichele, this project has the potential to open doors to research breakthroughs in several fields.
“Neuroscientists would like to understand how the brain can repair itself after a brain injury, for example. We want to understand how to make machines that can learn and adapt, by transferring principles from neuroscience to computers,” he says.
The “brain” will control the cyborg
The robot is beginning to take shape. The neuroscientists have cultivated a biological neural network from almost 100,000 neurons in a laboratory. Electrodes connect the neural network to a computer, and scientists can observe and analyse what the network is “saying” from the neural electrical signals. The researchers can also send signals back to the neural network – or stimulate it in different ways to “teach” it to behave in a certain way.
Photo: Kai T. Dragland, NTNU
Nichele and his IT colleagues are working on processing the data from the brain network and “translating” them into a language that the robot can understand. The researchers are learning to communicate with the biological material grown by the neuroscientists.
So far, the biological neural network has been communicating with an artificial neural network that controls the robot. Nichele hopes that he will eventually be able to remove the artificial brain network so that the biological neural network can “learn” directly within its working environment.
That will in turn enable the biological neural network to control the robot’s movements, so the robot can learn to adapt to different situations. This will lay the groundwork to build unique biological computers of living nerve cells that can learn over time.
Better understanding and treatment of central nervous system injuries
Modern biotechnology makes it relatively easy to build biological neural networks. For example, scientists can take cells from the skin of a rat – or human being – and transform them into neurons. Neural networks cultured today can live for up to a year under good conditions.
Researcher Ioanna Sandvig and PhD candidates Ola Huse Ramstad and Rosanne van de Wijdeven.
Researcher Ioanna Sandvig at NTNU’s Department of Neuromedicine and Movement Science does not so much think of these neural networks as a “brain”, but rather as a hierarchy of neurons. Even though the researchers are working with relatively simple networks, they can learn a lot about how the neurons communicate with each other and how connections are formed, maintained, and changed in neural networks. These relatively simple networks will be able to grow quite large and powerful over time.
The networks can be used for studying aspects of normal brain function, but also mechanisms behind disease and injury. Sandvig and her colleagues, including PhD candidate Ola Huse Ramstad, want to learn as much as possible about the neurons so that they can understand neuroplasticity, especially in the context of how the brain may be able to repair itself following injury. This knowledge could result in far better understanding and, hopefully, treatment options for patients with spinal cord injuries or stroke, or patients suffering from neurodegenerative diseases such as Parkinson’s or ALS.
“Although the neural networks that we’re culturing for this project are far from anything close to a brain, we’re still able to extract very important information by studying them. Even small advances may have great significance for these patients, she explains.
Lots of ideas
Thomas Rostrup Andersen from the Department of Engineering Cybernetics sits in a small office. He is one of four master’s students involved in the robot’s development. A cyborg prototype stands behind him. Evidently it’s resting.
“I’m working on creating a control module for the robot,” says Andersen, who will soon be submitting his thesis.
In fact, he is working on the robot’s core functions. Andersen is helping the various components talk to each other.
A camera records movements and faces, and a selfie arm can be raised and take pictures of people the cyborg encounters. The cyborg can be connected to a screen with a troll face that seemingly shows emotions. Mark Sagar, who also works with special effects for Hollywood, developed the troll face.
The researchers are testing out lots of ideas. Andersen and his successor are responsible for hanging out with the cyborg and making sure that the ideas work in practice and with each other.
A microelectrode array (MEA) chip containing neurons.
The cyborg is constantly evolving, and it hasn’t been a problem to recruit students for this project. NTNU master’s students and ten students from the Experts in Teamwork (EiT) course are participating in the work. Sverre Hendseth is the students’ adviser.
The Robot Operating System (ROS) is the open source program at the crux of the robotics part of the project. ROS is one of the standards used in the development of robots.
Biology and machine
PhD candidate and research assistant Martinius Knudsen in the Department of Engineering Cybernetics is the project coordinator for this part of the cyborg’s development. But his main job is to get the biological components of the cyborg to talk with the mechanical ones.
“Just developing a social robot is an ambitious goal in itself. Making biological neurons central to the project – in effect creating a cyborg – makes everything extra challenging. But by no means impossible,” says Knudsen.
Professor Gunnar Tufte and researcher Stefano Nichele at the Department of Computer Science.
Biological neurons fire electrical impulses. Machines can perceive and interpret these impulses. Microelectrode arrays, or MEAs, function as interfaces between the neurons and electronic circuitry, so that communication can go both ways between the robotic and biological parts. The underlying technology is complicated, but has to do with detecting extracellular electrical potentials.
Neurons require specific conditions to survive, like correct gas concentrations, nutrition and a sterile environment to avoid infection. So the most convenient way to keep the cultured neurons healthy is for them to continue to live at the neuroscience laboratory, at the Department of Neuroscience and Movement Science, at NTNU’s St Olavs campus, and to transmit the electrical impulses wirelessly between the neurons and the robot, which is located at Gløshaugen. It’s a little like having your brain somewhere else than in your body. Although the neurons hardly constitute a brain at this point.
The neurons “should probably be regarded more as a biological processing unit. The nerve cell network consists of about 100,000 neurons, which is still quite a way off from the 86 billion in the adult human brain,” says Knudsen.
But some simple life forms can function with that few – which brings up the ethical issues.
Is this the future we want?
Heidrun Åm is social scientist and researcher at NTNU’s Department of Interdisciplinary Studies of Culture. She thinks it’s important to be prepared for future technologies. What do we want them to do?
Åm does not think that NTNU’s combination of robot and biological cells poses a danger per se. But future technology will be far more advanced, so it’s about being prepared, she says.
In general, Åm considers it important to know what we are doing in order to ensure developments that are democratic and inclusive.
“We need an overview of the choices that are made in projects like these and the impact they can have on society. Only in this way can we make informed choices about whether this is the future we want,” says Åm.
It is important to include social scientists. Could the research have adverse effects? What’s my role in it all? Who takes responsibility if something goes wrong? We need to understand and regulate scientific development, so that it does not threaten the basic values of society.
What are scientists and engineers doing to ensure that most people can trust research developments to benefit them? Scientists and the rest of the population could quickly end up on a collision course.
“You have to win people’s confidence. If you don’t, the artificial intelligence debate will end up like what’s going on with genetically modified foods,” she says.