Exploring the generalization of skills with extra robot arms for motor enhancement

A recent study published in Advanced Intelligent Systems shows that the brain can adapt to artificial third arms and can be used for simple tasks. This will live on the precision mechanism and surgeon’s dreams for people to skillfully use their third arm in the future.

Approximately 20 study participants are learning to use artificial arms in laboratory settings. The rudimentary limbs equipped with clamps at the ends are secured to the table next to the participants. Exhaust moves your arms forward. The inhalation moves backwards.

Participants practice performing a series of tasks, such as grabbing a block, pressing a button, or moving a cursor. For scientists running this project, the goal is to determine to the extent that the brain can learn to use robotic limbs in the same way as natural arms.

In previous studies, the EPFL team had already demonstrated that participants could control the virtual arm and point to objects with a simple robotic arm. This goes a step further to examine your ability to grasp.

The team led by Postdoc David Leal measured skills as mediocre as complex. Generalising tasks. “We do this automatically using natural limbs,” explains Silvestro Micera, the lead author of the study. “If a child learns to grasp a particular object, he or she does not need to learn to do so afterwards for other objects. The brain internalizes the principles of motor sequences and generalizes it for other objects.”

Multitasking is almost impossible

According to Micera, if the brain can generalize tasks with an artificial arm, this shows that it can be incorporated. In other words, it is effective to use it as an integral part of the body. “It’s a clue that suggests that the brain can really control the robotic limb,” he says.

This study shows that generalization is actually taking place. Participants initially practiced moving blocking as quickly as possible using natural and artificial arms simultaneously. In the second stage, we were able to manipulate other objects more quickly and accurately, with natural arms and robotic arms, compared to non-practice participants.

In other words, an efficient protocol that induces generalization in natural limbs had the same effect as a robot limb.

However, if the operations requested during the test phase are too far from the training phase, generalizations are less. This is especially true in the context of multitasking. For example, participants find it difficult to generalize objects gripped with their artificial arms when they need to use their hands to simultaneously type the keyboard.

According to Micera, this result suggests that generalization with artificial limbs is more difficult to achieve and may be limited to performing very similar tasks. And perhaps training wasn’t optimal either, he adds.

Highly accurate research is too invasive

For the time being, relatively few scientists are working on augmenting the human being with robotic limbs. In the US and Europe, only a handful of teams are studying subject matter, including the integration of artificial fingers. Nevertheless, the promises inherent to this approach are intriguing.

“We can imagine many occupations where additional limbs can be useful. For example, first responders, precision mechanisms, or surgeons will no longer need an assistant to hand them the instrument,” says Michela. However, he quickly points out that such applications are not yet real.

The main obstacle is reduction control. Even if it is improved, diaphragm control of the artificial arm remains rudimentary, far from the accuracy of the natural limbs. To clear this hurdle, invasive interfaces such as electrodes within the cortex could be a long-term solution for converting brain signals into executable commands on the arm.

But this is not possible at the moment. Therefore, Micera and his team are limiting to non-invasive devices.

But for Misera, the appeal of such work is less than that in futuristic scenarios with extended humans to better understand how to interface and construct new connections between the brain and its body.

“For me, that’s more of a neuroscientific question than anything,” he explains. “By a better understanding of how to improve and speed up training with artificial arms, you can gain useful insights for rehabilitation, such as patients who have been paralyzed after a stroke.”