A new study has demonstrated a non-invasive technique that allows people to control a robotic arm using only their minds. The research could eventually lead to practical, mind-controlled prosthetics for those who suffer from paralysis or neurogenerative diseases.
Using one’s mind to control external hardware requires a brain computer interface (BCI). There are different types of BCIs, but they can be broadly grouped into two categories: invasive and non-invasive. An example of an invasive BCI is an array of electrodes implanted in a patient’s brain to measure neural activity. In contrast, non-invasive BCIs don’t require any surgery.
The non-invasive BCI used in the robotic arm study was an electroencephalography (EEG) based BCI. This type of BCI uses electrodes at the surface of the scalp to record the brain’s electrical activity. In the study, the thirteen test subjects were fitted with a specialized EEG cap lined with 64 electrodes.
Wearing the EEG cap, the participants learned to control a cursor on a computer screen by imagining where they wanted it to go. They were instructed to imagine either a left hand, right hand, both hands, or a relaxation of both hands, to correspond to the left, right, up/forward, down/backward movement of the cursor.
Whenever you move, or think about moving, the neurons in your motor cortex produce electric currents. Think about a different movement, and you activate a new assortment of neurons. Knowing this, the researchers used signal processing and machine learning techniques to match the EEG readings with a subject’s intended movement. Within just a few training sessions, the subjects were able to effectively control the cursor using their thoughts.
The researchers then turned the subjects’ two-dimensional control of a cursor into three-dimensional control of the robotic arm, using a two-stage approach. For example, in a task where subjects were instructed to pick up an object with the arm, the first stage involved a 2D plane in a region above the object. The subjects would begin by moving the cursor/arm into position directly above the object.
Then, in the second stage, the subjects would guide the arm downwards in the third dimension in order to grasp the object. In this way, the number of degrees of freedom that the BCI needs to interpret is reduced, yet subjects can still interact with objects in 3D space.
The subjects performed multiple tasks with the robotic arm, from grasping random objects on a table to picking up the objects and moving them to a shelf. Considering they were controlling the arm completely with their imagination, the results were impressive: the average success rate for picking up objects in fixed locations was over 80 percent, and moving objects from the table to the shelf had an average success rate of over 70 percent.
“This is exciting as all subjects accomplished the tasks using a completely non-invasive technique,” said lead researcher Bin He from the University of Minnesota. “We see a big potential for this research to help people who are paralyzed or have neurodegenerative diseases to become more independent without a need for surgical implants.”
You can read the team’s research paper in Scientific Reports. To learn about a different type of BCI, read Playing Video Games with Direct Brain Stimulation.