Scientists develop real-life mind controllable robotic hand using AI

It may sound like science-fiction, but the researchers claim that the new technology would allow patients to control robotic body parts using their thoughts. By employing artificial intelligence and machine learning, the researchers at UMN have developed a portable neuroprosthetic hand. The robotic hand comes equipped with a nerve implant linked to the peripheral nerve in a patient’s arm. 

Explaining the significance of their neuroprosthetic innovation, project collaborator and UMN neuroscientist Edward Keefer said, “We are well along the way toward allowing upper limb amputees at least, and other people in the future, to have totally natural and intuitive control of their prosthetic devices.” 

The neuroprosthetic hand is different from your regular prosthetic limbs

The prosthetic body parts currently available on the market detect shoulder, chest, or muscle movement. They have sensors to recognize signals in specific regions of the human body. Therefore, every time a patient wants to move his hand, he is required to trigger his body muscles. Adapting to such muscle-driven limb movement is not easy for patients, and many such devices are not suitable for physically weak individuals.

Some advanced and efficient muscle-sensitive prosthetics come with complex wiring and other arrangements that make them difficult to use. The amputees have to go through a lot of training to adjust to such devices, which often increases frustration and stress. Now imagine a device that starts working immediately, is less invasive, requires no training, no muscle activation, and no complex setup. 

The neuroprosthetic arm enables the patients to move their arms simply at the will of their minds. It is an efficient, easy to use, and a lot more intuitive alternative to any commercial prosthetic system available.

Researcher and one of the authors of the study, Jules Anh Tuan Nguyen, said, “With other commercial prosthetic systems, when amputees want to move a finger, they don’t actually think about moving a finger. They’re trying to activate the muscles in their arm since that’s what the system reads. Because of that, these systems require a lot of learning and practice. For our technology, because we interpret the nerve signal directly, it knows the patient’s intention. If they want to move a finger, all they have to do is think about moving that finger.”

What makes the mind-controllable robotic arm possible?

According to the researchers, the primary difference between traditional muscle-sensitive prosthetics and robotic hand is the latter’s ability to understand nerve signals. Due to the integration of artificial intelligence in their device, the implant can function as a neural decoder (a device capable of understanding electrical neural signals). 

Hand movement in humans is achieved by the action of the median nerve, a type of peripheral nerve which is composed of thousands of axons. When a patient thinks of moving his hand, his brain signals the peripheral nerves. These signals are detected by the neural decoder linked to the nerve, which then directs the robotic arm to function accordingly. The neural decoder uses machine learning principles to interpret nerve signals.

The machine learning-driven robotic arm setup currently requires a wired connection between skin and an external AI interface. However, the researchers hope that the implant could be linked remotely to any computer in the future. This would enable amputees to perform limb movement easily and allow any human being to control the different smart devices around him (such as smartphones) with his brain.

One of the leading developers of the neural chip technology and study author Zhi Yang believes, “this technology was designed for amputees for sure, but if you talk about its true potential, this could be applicable to all of us.”

The study highlighting the neuroprosthetic hand research is published in the Journal of Neural Engineering.