A virtual drone was steered through an obstacle course by imagining moving a finger.
Wilsey et al.
A paralyzed man with electrodes implanted in his brain can pilot a virtual drone through an obstacle course just by imagining moving his fingers. His brain signals are interpreted by an AI model and used to control a simulated drone.
Research on brain-computer interfaces (BCI) has made great progress in recent years, allowing people with paralysis to write speech on a computer by precisely controlling a mouse cursor or imagining writing words with a pen. It became. However, so far it has not yet shown much promise in complex applications with multiple inputs.
now, Matthew Wilsey Researchers at the University of Michigan created an algorithm that allows users to trigger four discrete signals by imagining moving their fingers and thumbs.
The anonymous man who tried the technique is a quadriplegic due to a spinal cord injury. He was already fitted with Blackrock Neurotech's BCI, which consists of 192 electrodes implanted in the area of ​​the brain that controls hand movements.
An AI model was used to map the complex neural signals received by the electrodes onto the user's thoughts. Participants learned how to think about moving the first two fingers of one hand to generate electrical signals that can be made stronger or weaker. Another signal was generated by the next two fingers, and another two by the thumb.
These are enough to allow the user to control the virtual drone with just their head, and with practice they will be able to expertly maneuver it through obstacle courses. Wilsey said the experiment could have been done using a real drone, but was done virtually for simplicity and safety.
“The goal of building a quadcopter was largely shared by our lab and the participants,” Wilsey says. “For him, it was a kind of dream come true that he thought was lost after he got injured. He had a passion and a dream to fly. He felt so empowered and capable. He instructed us to take a video and send it to a friend.
Although the results are impressive, Willsey says there is still much work to be done before BCIs can be reliably used for complex tasks. First, AI is required to interpret the signals from the electrodes, but this depends on individual training for each user. Second, this training must be repeated over time as function declines. This could be due to slight misalignment of the electrodes in the brain or changes in the brain itself.
topic:
Source: www.newscientist.com
