Operating a robotic arm, prosthetic hand, or rehabilitation device is more challenging than it may seem. Picking up an egg requires precise force—too little and it slips, too much and it cracks.
For prosthetic users or stroke patients, this precision is difficult due to limited or absent visual and tactile feedback. Without sufficient sensory information, controlling movements accurately becomes much more difficult.
To address this, researchers have added cues like vibrations, sounds, and visual signals to replace missing sensations. Augmented sensory feedback can improve performance, but it requires extra equipment and cannot fully replicate natural limb sensation.
Seeking a simpler solution, Pierre Vassiliadis and Friedhelm Hummel at EPFL’s Neuro-X Institute, with Silvestro Micera and Solaiman Shokur, explored a different approach. Instead of restoring lost sensations, they tested whether the brain could learn better from feedback tied directly to successful actions.
Continuous feedback
“Most training methods only tell users whether they succeeded after completing a movement,” Vassiliadis explains. “However, a final score or success notification does not indicate which part of a complex movement caused the error.”
To address this limitation, researchers at EPFL developed a system that delivers performance feedback while the movement is taking place. Across five studies with 106 participants (including 18 stroke patients), participants tracked a moving target for seven seconds using a cursor controlled by a force sensor or biceps activation.
During the task, the target changed color in real time based on performance—green for success and red for errors. As performance improved, the system automatically adjusted the feedback threshold to maintain an appropriate level of challenge and keep the information useful. In comparison experiments, the color changes were random, and participants were instructed to disregard them.
With fewer than 20 color-based trials, participants rapidly improved motor control, and the gains persisted even after feedback was removed.
People Respond Differently To Training and Feedback
The color-based feedback proved most effective when other feedback was restricted. When participants saw the cursor only about one-third of the time, performance gains were roughly three times higher than with full visual feedback.
A similar result appeared in another experiment using a muscle-activity control system: reducing artificial touch feedback further amplified the benefits.
Stroke patients also showed improvements in low-vision conditions, although these gains did not last after training ended. Researchers attribute this to the brief training period and possible differences in how motor memories are formed following brain injury.
People varied in how strongly they benefited from the training. Higher reward sensitivity—a trait linked to the brain’s reward system—was associated with greater gains in both healthy participants and stroke patients, suggesting it may be possible to predict who will respond best to this approach.
The researchers also examined how information was exchanged between users and the interface. They found that real-time reinforcement helped make up for reduced moment-to-moment motor corrections when sensory input was limited. Instead of pushing users to try new strategies after errors, the color feedback reinforced and strengthened movements that were already successful.
According to Vassiliadis, the simplicity of the method means it could be easily integrated into existing prosthetic, rehabilitation, and human–machine interface systems with minimal added cost. By leveraging the brain’s natural reward-learning mechanisms, real-time feedback could provide a scalable way to make motor training faster, easier, and more effective.
Read the original article on: Tech Xplore
Read more: Robot dogs will be deployed to patrol stadiums during the World Cup