This story is over 5 years old.

Drones Are Flying Under Thought-Control

New research demonstrates that it's pretty easy, too.
June 4, 2013, 11:10pm

First of all, of course we can control drones with thoughts. It’s a splashy headline conjuring all sorts of fun sci-fi imagery, but drone technology and BCI (brain-computer interface) technology have been racing forward over the past several years on more or less parallel tracks. The quadcopter, the current standard-issue runabout drone, is perfectly built for control-based experiments, while the $750 Emotiv EEG "neuroheadset" is already being used to control a motorized skateboard out-of-the-box. It’s somewhat of an understatement, but, fundamentally, a thought-controlled drone is just a matter of plugging the two technologies into each other.


It’s when you get to the actual, evolution-gifted brains that the question comes in: are humans capable of thought-control in three-space?

A study out today from the University of Minnesota and published in the Journal of Neural Engineering demonstrates that at least five humans/subjects are up to the task of steering a quadcopter through some balloon hoops in a gymnasium using just thoughts. The task is conceptually simple enough, but the team didn’t know if it would work at all. “I was not sure we were even able to do it,” Professor Bin He, the study’s lead author, recalls in a phone interview from his office. “I was certain we would be able to do it in some way, but I was surprised by the excellent performance we were able to accomplish with this group of subjects.”

The task feels super-mundane—jumping through hoops, literally—but it proves an important possibility. That possibility is, however, less a world comprised of brains encased in warehouses connected to a physical, day-to-day society of helicopters—helicopters going to jobs, being romantic, running nations, etc.—than it is of people suffering from neurological disabilities or paralysis being able to navigate a three-dimensional environment using EEG. Whoa awesome headline or not, this is vital research being done to improve the lives of humans currently living in the non-sci-fi world.

There are a few details about the study needed for perspective. The first is that participants using BCI performed about one-quarter as good as the control group, which used regular keyboard interfaces. This, Professor He says, is impressive. “[It] means that we did a fairly good job. Think about the keyboard and how fast you can move, and then that we only trained the subjects [with brain-control] for 10 or 12 hours.”


While more training should improve this performance level, getting to the point that thoughts approach the same level of control as keyboards remains unlikely. That is, thought-control probably isn’t going to take over from physical interfaces anytime soon. “If you happen to use even the joystick to control this robot, I believe that it would not be that easy to accomplish 100 percent,” He says.

Professor Bin He with experiment

The second detail, and what should explain the difference in performances, is that interface isn’t so much pilots just looking at different places on a screen and, thus, directing the drone in that direction. That would be nice, but BCI isn’t quite set to deliver so much just yet. Instead, users are trained to just think about moving either their right or left hands, which then causes the drone to move in that direction. Imagining both hands being used at once makes the drone go up. That’s it—velocity was preset. So, the BCI is only sort of/not really removing the mediation of a keyboard or joystick.

That’s indeed a bit less fun than getting to be a robot bird, but still. Just watch the video above. The next step actually has nothing to do with flying—remember: this isn’t actually about being airborne—and instead will focus on robot arms. Professor He is quick to emphasize that his interest is clinical—helping people—but notes that thought-control in three-space has potential applications in driving and, naturally, video games.

Reach this writer at