Even robots make mistakes sometimes. That’s why researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) have built a system while allows robots to be corrected through thought and hand gestures.
The system monitors brain activity, determining if a person has noticed an error in the machine’s work. If an error is detected, the system reverts over to human control. From that point, all it takes is a flick of the wrist to get the robot back on the right course.
“This work combining EEG and EMG feedback enables natural human-robot interactions for a broader set of applications that we’ve been able to do before using only EEG feedback. By including muscle feedback, we can use gestures to command the robot spatially, with much more nuance and specificity,” says CSAIL Director Daniela Rus, who supervised the work, in a press statement.
EEG refers to electroencephalography, a type of biofeedback which uses real-time displays of brain activity to each self-regulation in the brain. EMG feedback refers to electromyography, which is the recording of the electrical activity of muscle tissue.
Earlier brain recognition systems required people to think in highly specific ways to achieve EEG or EMG recognition. What Rus’ team realized is that when the human brain recognizes an error, it automatically releases a very specific signal all on its own. These signals are called error-related potentials (ErrPs). When the robotic system notices an ErrP signal in the human brain, it turns the robot over to human control.
“What’s great about this approach is that there’s no need to train users to think in a prescribed way. The machine adapts to you, and not the other way around,” says Ph.D. candidate Joseph DelPreto, who also worked on the project.
The team used a worker bot called “Baxter”, from Rethink Robotics, for testing. Using the MIT biofeedback system, the robot was able to improve its accuracy from 70 percent to 97%.
There are numerous potential uses for such a technology. It could continue on in Baxter’s path and help automate factories even more, with just a few supervisors overlooking errors and letting their brain chemicals act as correctives on the machines. The researchers also emphasize that it could be helpful for the elderly or those with limited mobility.
“We’d like to move away from a world where people have to adapt to the constraints of machines. Approaches like this show that it’s very much possible to develop robotic systems that are a more natural and intuitive extension of us,” says Rus.
MIT’s CSAIL program has a focus on robots that can fit into everyday life as it currently stands. With researchers from the University of Toronto, for example, they’ve been training AI to learn from The Sims so it can get better at doing human chores.
This article was originally published in Popular Mechanics