Imagine that you have a robot which mistakes you could correct simply by using the power of your thoughts. A robot that will be more like a natural extension of us, capable of learning human language. That is exactly what researchers have worked on. Creating a robot that is able to changes its actions almost immediately after person observes it making a mistake.
A team from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) and Boston have developed a robot that takes its tips directly from human brain waves, allowing the feedback system that lets people correct robot mistakes just by their thoughts. The system uses electroencephalography (EEG) to measure a person’s brain signals as they watch a robot work. When it detects a signal suggesting the person has witnessed a mistake, it modifies the robot’s course. Thanks to the new algorithms that monitor brain activity for specific signals, a person can give a command to the robot without spoken word or pushed button.
For this research purposes, the team used a humanoid named “Baxter” from Rethink Robotics, the company led by former CSAIL director and iRobot co-founder Rodney Brooks. They took five volunteers, attached with EEG headset and observed robot behavior on LED light. For each test one LED light was randomly assigned as correct one. Every time when Baxter chose the wrong LED light, the system recognized it and generate brain signals called “error potentials”. The same method is used for a test where volunteers watched Baxter sort reels of wire and paint bottles into different boxes.
An advantage of using error potentials is that people don’t need to “think” about specific words or movements to generate commands as in some previous work in EEG-controlled robotics. Basically, there is no need for any training to use the system. CSAIL Director Daniela Rus explains it: “As you watch the robot, all you have to do is mentally agree or disagree with what it is doing. You don’t have to train yourself to think in a certain way, the machine adapts to you, and not the other way around”. In this way, the research team wanted to things become more natural.
The accuracy rate of conducted tests is 70 percent and the team plans to increase it by at least 90 percent. The findings are due to be presented at the IEEE International Conference on Robotics and Automation in Singapore in May.
CSAIL published a video where they presented a feedback system developed at MIT which enables human operators to correct a robot’s choice in real-time using only brain signals.