Research on human-robot interaction is of interest to many universities. For this team of MIT, the emphasis is on the correction of the machine by a human operator equipped with an EEG helmet equipped with 48 sensors.
The robot can thus recognize if it has carried out a bad manipulation by detecting a signal of the device called "error potential". The latter is clearly distinguished from other cerebral waves by its intensity and precision. It also reflects the reaction of an individual who notices a fault.
Establishing Mental Control
The system has proven itself in a demo video where a Rethink industrial robot is sorting between wire reels and aerosols. If he sometimes deceives himself, it is only temporary, since he corrects his action knowing that he made a mistake. For now, the robot has chosen between two options, but the team led by Daniela Rus wants to further improve the system. It is also looking at an application in stand-alone technologies. In the case of a car, the passenger could alert the system when a hazard occurs. For his part, Joseph DelPreto, PhD student at MIT, thinks industry remains the main concern.
0 comments:
Post a Comment