Michael Mehringer is paralyzed from the neck down. Together with a team of neuroscientists, neurosurgeons, robotics experts, and AI researchers, he is working to gain more autonomy through a brain-computer interface.
Astrid Eckert / TUM The arm on the computer screen looks like something out of a video game. The hand, joints, and arm bones are built from orange polygons. When the animation starts, the arm extends. After a few seconds, everything returns to its starting position. Michael Mehringer watches the screen intently. "Excellent! And again," says Melissa Zavaglia. The animation starts again. Mehringer remains focused.
The 26-year-old has been paralyzed from the neck down since he was in a serious motorcycle accident about ten years ago. He can only reproduce the animated arm's movements mentally. "Still, after the sessions, I can always feel how much work I've done," Mehringer says. "My body hasn't been through movements like that in years." The exercises with the animated arm are part of a research project aimed at providing new insights into how the brain works. Beyond that, the researchers want to enable him to control computers-and even a robotic arm-using the power of the mind.
Astrid Eckert / TUM A small black box, about the size of a matchbox, sits on the back of Mehringer's head. Two cables lead from it to a second computer. Its screen shows a grid of 256 rectangles. The base color of the grid is blue, sometimes lighter, sometimes darker, always moving like a pixelated image of a water surface. At various points, the blue is interrupted by other colors: green, yellow, red, up to muted maroon. These colors are constantly shifting, too. Each rectangle represents one of the 256 electrodes measuring Michael Mehringer's brain activity. If one turns dark red, activity in the cells around that sensor is especially high. Which brain cells become active depends on the task.
If Mehringer imagines the arm movement, the grid lights up green, yellow, and red in the upper left. In another, later task, it's in the bottom right. Such a detailed view into the brain is only possible because Mehringer agreed to a far-reaching procedure: in summer 2025, four electrode arrays were implanted directly into his brain during an operation lasting more than five hours. Each array is about 5 by 5 millimeters and carries 64 needle-like microelectrodes.
"The greatest challenge was implanting the electrodes with absolute precision. That's the only way to obtain accurate recordings and measure brain signals reliably," explains Professor Bernhard Meyer, Director of the Department of Neurosurgery at TUM University Hospital. Professor Meyer and his team planned and prepared the procedure for years. They wanted to be sure they would hit the optimal point in the brain area responsible for planning and executing complex movements.
"With this operation, a microelectrode-based brain-computer interface was used in a person with quadriplegia for the first time in Europe," says Simon Jacob, Professor of Translational Neurotechnology. This makes TUM the first academic institution in Europe to have already implanted two such devices. In 2022, the team had implanted one in a stroke patient with a language disorder, which made it possible to map language processing in the healthy right hemisphere of her brain.
Artificial Intelligence
Artificial intelligence is shaping our working lives, research, and the world around us. Learn how we are contributing to this progress by developing innovative AI methods and applications, ranging from robotics to machine learning.
Andreas Heddergott / TUM Such milestones are only possible through dedicated teamwork and collaboration across scientific disciplines. "Artificial Intelligence for Neuro Deficits" is the name of the research project funded by Germany's federal research ministry, combining experience from brain surgery with the knowledge of neuroscientists and expertise from robotics.
Translating brain signals into movement
Twice a week, a transport service takes Mehringer to the Neuro-Head Center at TUM University Hospital. Today, Dr. Melissa Zavaglia, who leads the project at the Munich Institute of Robotics and Machine Intelligence, is there with two doctoral researchers. During the training, neurologist and neuroscientist Simon Jacob will stop by. The neurosurgery team is always available to provide medical care and monitor the interface.
Astrid Eckert / TUM A quick health check, and some relaxed conversation while the computer connection is prepared and disinfected. Mehringer takes a sip of coffee through a straw, and then they begin. Not many words are exchanged during the training session. Everyone knows their tasks. Now and then, the researchers ask Mehringer whether everything is working for him. It is.
The TUM research team uses exercises like the animated arm to train AI algorithms. The aim is for the system to recognize the relationship between the neuronal signals and the movement Mehringer wants to perform. At first, the decoded brain signals will be used to control a cursor on a screen.
Then, the researchers hope, Mehringer can gradually learn to move a robotic arm and use it to grasp objects. "Rather than expecting humans to conform to and learn how to operate robotic systems, the focus is on designing systems that recognize human intent," Zavaglia says.
Michael Mehringer is central to the team. If anything is delayed with him, all other appointments shift as well. The training sessions are organized so they're manageable for him. "Before we started this, I mostly pictured people in white lab coats-not something as laid-back as this," he says.
Along with the new possibilities they offer, brain-computer interfaces also raise ethical questions. Marcello Ienca , Professor of AI and Neuroscience Ethics at TUM, is examining these issues in close collaboration with the team. He has served on expert panels at the OECD and UNESCO and is the designated president of the International Neuroethics Society.
No one can yet say whether a thought-controlled robotic arm will ultimately be suitable for everyday use. "It's important for participants in our studies to understand that at its core, this is about research, not a cure. Research isn't as predictable as swallowing a headache pill that has been tried and tested for countless years," Jacob says.
Michael Mehringer is aware of this. At the same time, he is convinced that he has made the right decision: "I'm proud that I can help move research forward. Of course I'd like to have full independence again, but I'm a realist. If I could drink-or even eat-whenever I wanted with the robotic arm, that would be a very big step forward for me and a lot of freedom regained."
This article was published in the third issue of TUM Magazine.
