News

Australian Army “Mind Controls” Robot Dog With Brain-Machine Interface

March 30, 2023 by Aaron Carman

In the near future, robots may take little more than a thought to control.

Breaking down the communication barrier between humans and computers, University of Technology Sydney (UTS) researchers have developed a new biosensor technology that allows users to control devices with their own brain waves. The technology itself unifies the well-developed knowledge behind silicon manufacturing with graphene material to deliver a more conductive and better-performing sensor.

 

The technology developed by UTS

The technology developed by UTS was demonstrated by the Australian Army, where soldiers controlled a quadruped robot using the UTS brain-machine interface. Image used courtesy of the Australian Army

 

Brain-machine interfaces (BMIs) are not a new idea, with companies like Elon Musk’s Neuralink hoping to develop an acceptable method for users to control devices such as prosthetics. The solution developed by UTS, however, represents an ultra-compact, on-body device that may offer an effective method of scaling BMIs to the point of universal adoption.

In order to give readers an overview of the benefits and limitations of this technology, this article delves into the BMI reported by UTS researchers and the applications for which it may be suited.

 

A Graphene-based Dry Sensor Measures Brain Waves

If you want to know what’s going on beneath your skull, a craniotomy will give you the clearest view. Now, however, there are much less invasive means of bringing your inner thoughts to the outer world. Electroencephalography (EEG) electrodes represent just one way of reading brain waves and traditionally rely on “wet” sensors using gel to reduce the contact resistance between the skin and the sensor. These sensors, however, are not optimal for continual use since the gel can present a myriad of problems, such as irritation or drying.

Dry sensors, if able to perform as well as their wet counterparts, would make it much easier to integrate BMIs into daily life. As such, UTS researchers leveraged their expertise with graphene to develop their own dry BMI with improved performance.

 

Fabrication used by UTS researchers to develop their BMI device

Fabrication used by UTS researchers to develop their BMI device. Image used courtesy of Applied Nano Materials

 

Several patterns were tested, but ultimately, the best pattern for a dry BMI sensor to be used in a hairy region of the scalp used hexagonal pillars of graphene. Although this sensor doesn’t provide the lowest contact impedance in controlled tests, it does provide the best performance in a practical sensor considering the head’s hair and curvature.

 

Real-time “Mind Control”

In order to evaluate the sensor’s performance in a realistic sensing situation, two independent tests gave quantitative and qualitative data on the UTS BMI’s ability to “read minds.” The first of these evaluated is the signal-to-noise ratio (SNR) of the sensor compared to the gold-standard EEG electrodes. Researchers found that the dry sensor, while unable to achieve the 30 dB SNR of the wet sensor, exhibited an SNR up to 25 dB, bringing it within striking distance of the gold standard.

The UTS group also found that, in addition to the lower SNR, the placement of the dry sensor introduces considerable variability in the SNR measurement, with values as low as 5 dB reported. While this is an issue that must be addressed, it may provide room for improvement in future systems using optimization.

 

The results from the laboratory SNR test

The results from the laboratory SNR test show that, although the dry sensor has a higher signal level, the reduction in noise in the wet sensor still makes it a more reliable solution for the time being. Image used courtesy of Applied Nano Materials

 

Finally, to evaluate the current BMI’s ability to control a practical device, UTS partnered with the Australian Army to integrate the newly developed BMI with the army’s quadruped robot. In the tests, a single operator was able to give the robot up to nine commands in a two-second window. Sergeant Damian Robinson called the process “very intuitive.” So, while the UTS sensor may not be widely deployed in the near future, the preliminary results show that it has the potential to make human-robot communication much simpler.

 

Hold That Thought

While the technology presented by UTS and demonstrated by the Australian Army is exciting and futuristic, it relies heavily on on-body sensors and is still susceptible to inaccuracies that scale with the complexity and speed of control. In that sense, scenarios such as military conflicts, high-precision manufacturing, or prosthetic development would benefit greatly from increased human-robot synergy, but BMIs are not currently a universal interface solution.

While the UTS BMI telepathically gave commands to the Australian Army’s robotic dog, it will be interesting to see the ultimate complexity, precision, and speed with which commands can be sent, especially compared to endovascular sensors. These factors seem to be the current limiting factor for BMIs, but as the technology evolves and new developments like the ones from UTS are made, keyboards and controllers may one day become a thing of the past.