To expand human-robot collaboration in the manufacturing space, researchers from Loughborough University in the U.K. are training artificial intelligence (AI) to detect human intention, thereby improving human and robot interactions.

Although robots are designed to make assorted tasks easier on humans, humans are still necessary for certain duties in the manufacturing space that demand flexibility, adaptability and dexterity. As such the Loughborough University team is training robots to detect human arm movement intention prior to human movement actually being articulated.

Source: Loughborough UniversitySource: Loughborough University

To accomplish this, the team trained the AI to recognize pre-movement patterns captured by an electroencephalogram (EEG), which is technology that records brain activity.

In the lab, the researchers tested the AI with the help of eight participants seated in front of computer screens. According to researchers, the computers would randomly generate a letter from A to Z on the screen while the test participants would press the corresponding key on the keyboard. Meanwhile, the AI system predicted which arm the participants would move according to the EEG data and motion sensors confirmed the intention.

The data demonstrated that the AI system detected when the participant was preparing to move his or her arm up to 513 milliseconds (ms) before they moved it. On average, the AI system detected the movement roughly 300 ms ahead of execution.

The study, EEG based arm movement intention recognition towards enhanced safety in symbiotic Human-Robot Collaboration, was published in the journal Robotics and Computer-Integrated Manufacturing.

To contact the author of this article, email mdonlon@globalspec.com