Humans have finally developed a way to communicate their displeasure with robots without kicking them.
Researchers at MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL), with Boston University, developed a feedback system for humans to mentally change a robot's actions by simply watching and wordlessly judging.
The task at hand was a simple sorting test. A robot called "Baxter," from Rethink Robotics, set out to organize paint and wires into appropriately-labeled bins. When Baxter's hand headed for an incorrect bin — it didn't know the difference between the paint or wire in its hand — the supervising human sent faint "error-related potential" brain signals through an EEG monitor to Baxter (which then smirks and blushes). The robot knows it's on the correct track by picking up on these faint blips of "oh, that's not right" brain waves. The algorithm does all of this in a matter of 10 to 30 milliseconds.
The researchers are hoping this technology will spur new methods for humans and robots to have conversations in more than just binary yes-or-no ways. Complex multiple-choice problems could be next, used in prostheses, non-verbal communication, machine learning, and near-genuine conversations between man and machine.
It's not the first time EEG and brain waves have been used to communicate with and control machines. People are flying drones with their minds, and even getting into mind-controlled cockpits to fly themselves around. But the simplicity of MIT CSAIL's new research makes it more accessible to the public, no new language or mental training required.
In the not-too-distant future, crews of robotic factory workers could be managed with telekinetic-like control from a human supervisor. Once the robots start talking back, will they ask us to lay off the kicking?