AI enables a person with paraplegia to control a robotic arm with their thoughts

Four people, three healthy volunteers and one with cerebral palsy, have managed to control external devices using their brain activity thanks to a brain-computer interface (BCI) enhanced with artificial intelligence (AI).

In the study, published in Nature Machine Intelligence , a person with paraplegia was able to control an on-screen cursor almost four times better thanks to the support of AI algorithms. He was also able to complete tasks with a robotic arm that would have been impossible without this assistance.

The developed system uses a shared control model: the BCI records brain activity and the AI ​​interprets the user's intention, guiding movement more fluidly and precisely.

This noninvasive brain-computer interface system can read brain activity through electrodes and use machine learning to improve movement control. The interface featured two AI copilots: one to help guide the computer's cursor and another to assist with the robotic arm's tasks using visual information.

When testing the interface with AI co-pilots, the participant with leg paralysis due to a spinal cord injury achieved 3.9 times better performance in controlling the computer cursor than without the assistance of the AI ​​co-pilot.

Participants without paralysis experienced 2.1 times better performance after activating the AI. Similarly, the

A participant with paralysis was able to control a robotic arm to move colored blocks to specific targets, which was previously not possible without AI support.

According to the team from the University of California (USA), artificial intelligence can act as a " copilot " and significantly improve performance.

"Many patients with paralysis maintain intact brain activity. Brain-computer interfaces transform this activity into commands to operate robotic arms, cursors, or wheelchairs. The novelty of this work is that AI helps deduce the user's objective and compensate for incomplete or noisy signals, which improves the experience and efficiency," Eduardo Fernández, director of the Bioengineering Institute at Miguel Hernández University and of the Biomedical Neuroengineering group at CIBER-BBN , told the Science Media Centre .

The study is still preliminary: only three healthy volunteers and one person with a spinal cord injury participated, and the tests do not simulate everyday activities. Still, the results point to a more intuitive and functional framework for the development of future BCIs.

Researchers say this shared control model could make brain-computer interfaces more practical and effective for everyday use, and as AI systems improve, they could help users perform more complex tasks more easily.