Leggi in italiano

In the experimental set-up, a chest belt monitores the movements of the diaphragm, so that participants could use the way they breathe to control the elongation of the arm. Credit: Alain Herzog/EPFL.

Researchers have developed a robotic arm that can be controlled using gaze and the movement of the diaphragm during breathing. The arm, which is designed to be used as an extra limb rather than a replacement, has proved to be easy to learn to use and it can be used while conducting other activities1.

Extra robotic arms are designed to enhance the capabilities of healthy individuals and could be used to assist workers or older people or perhaps provide surgeons with an extra hand. But controlling an extra arm cannot rely on strategies developed for those with missing limbs, such as the use of residual nerves in an amputee’s arm to control the prosthesis.

“We have had to deal with what we call the neural allocation problem, which involves finding ways of channeling motor commands and sensory information to and from the device without hindering the control of biological limbs,” explains Silvestro Micera who heads biorobotics laboratories at Scuola Superiore Sant’Anna in Pisa and the Swiss Federal Institute of Technology in Lausanne.

In the study published in Science Robotics, the study team first tested the control system in a virtual environment. Participants wore virtual reality headsets that tracked gaze to orientate a virtual third arm, which appeared to emerge from the middle of the chest and had two thumbs to avoid being associated with the left or right hand. A chest belt monitored the movements of the diaphragm, so that participants could use the way they breathe to control the elongation of the arm. When doing a simple task of reaching out to touch a virtual object, the participants had a success rate of about 60% using the virtual arm both in isolation and in combination with their own arms. When researchers asked the participants to execute the task while counting out loud or spotting visual clues in their peripheral view, there was no significant decline in success rate.

The researchers then developed a simplified physical version of the arm that could rotate on a plane and elongate and weighed less than two kilograms. “We did not integrate any sensory feedback in the physical test, since we learned in the virtual environment that this did not help execute the task,” Micera says. Participants who had previously learned to use the extra virtual arm had a success rate of about 90%, whereas participants who had not used the virtual arm were nearer to 70% successful. The team is now working on a more complex version of the robotic arm that will be able to exert more pressure and move in three dimensions.