A New Non-Invasive Brain-Machine Interface Offers Thought-Based Robot Control with High Accuracy

Researchers from the University of Technology Sydney (UTS) have developed a non-invasive sensor which, they say, allows for the control of robots through thought alone — and they've already partnered with the Australian Army to put the technology to use.

"The hands-free, voice-free technology works outside laboratory settings, anytime, anywhere. It makes interfaces such as consoles, keyboards, touchscreens and hand-gesture recognition redundant," boasts Francesca Iacopi, UTS professor and corresponding author on a new paper detailing the technology. "By using cutting edge graphene material, combined with silicon, we were able to overcome issues of corrosion, durability and skin contact resistance, to develop the wearable dry sensors."

Brain-machine interface (BMIs), also known as brain-computer interfaces (BCI), aren't a new concept, and this isn't the first time they've been demonstrated — but previous efforts at full mind control of complex robotics have typically relied on invasive sensors inserted into the user's brain. The UTS system, by contrast, is simply positioned in a hexagonal grid on the back of the user's scalp — yet can withstand harsh environments, its creators claim.

"Our technology can issue at least nine commands in two seconds. This means we have nine different kinds of commands and the operator can select one from those nine within that time period," Chin-Teng Lin, professor and co-author Chin-Teng Lin explains. "We have also explored how to minimise noise from the body and environment to get a clearer signal from an operator’s brain."

Those nine commands are displayed to the operator using an augmented reality display. Concentrate on one of the nine displayed commands, and it's sent to the connected robot automatically. In field tests with the Australian Army, in which BMI-equipped soldiers were told to use the sensor to operate a Ghost Robotics quadrupedal land drone, the system worked with a 94 per cent accuracy rate.

The team's work has been published under open-access terms in the journal ACS Applied Nanomaterials.