It’s hard to tell this story without making a Skynet joke, but here goes: Researchers have developed a prototype robot that can take control of a human partner’s arm and use it to complete tasks. Thanks to electrodes strapped to the human’s arm and a technique called functional electrical stimulation (FES), the robot can send signals that will cause the arm to move and the hand to open or close. From the perspective of the human, these movements are entirely involuntary. From the perspective of the robot, it is controlling a second serial robot.
Here’s a video demonstration made by Bruno Vilhena Adorno at the Montpellier Laboratory of Informatics, Robotics, and Microelectronics (known by its French acronym LIRMM):So why are they doing this? Remember, the human’s arm is only moving because the robot told it to. It’s a loss of control that Adorno admits can feel strange and uncomfortable. Consider the problem of a person with limited mobility who, perhaps due to paralysis, doesn’t have full control of their limbs. Generally, the helper robot would perform tasks for the human, but there are some tasks that might require greater precision or a greater range of movement than the robot has. Rather than throw engineering resources at making increasingly perfect robots, FES enables increased precision by giving the human half of the equation some ability to contribute to the action.
Instead of a a highly capable robot assisting a nearly inert partner, Adorno envisions situations where the two players are acceptably capable and cooperating. This permits a degree of precision greater than the sum of its parts. While the demonstration involves a hoop and ball, imagine the problem of getting a coat onto someone with no mobility versus someone able to at least hold their arms out.
Beyond assisting people with reduced mobility, Adorno says that there could be applications for robot-human interaction designed to enhance a person’s capabilities. For example, FES could act as a set of artificial reflexes that a robot armed with specialized sensors could use to protect people from otherwise undetected dangers. Adorno says that this kind of application might be better suited to exoskeletons but that the cooperative techniques would still be applied.
It’s a neat approach and it points to a whole host of wider questions about how to best manage the division of responsibility between humans and our helpers as we prepare to live in society with machines. There will be situations where we want to be able to override the actions of our robots, but as this demonstration shows, there will be situations where we want the robot to have the final say. Though Adorno is careful to emphasize that this research is intended to assist impaired individuals, the demonstration makes it clear that it could be applied to anyone, for better or worse. As for the public response, Adorno says that it’s generally been positive, though the public "tends to have a lot of imagination."
"It is quite common to hear things like 'wow, the next step is Skynet controlling all humans, we’re doomed,'" he says, "Of course, in general people are just kidding when they say that."