New System that Links Your Mind to Robot to Help Prevent Mistakes

How do you prevent your robot from carrying out a task that could turn out to be disastrous? You could tell it verbally, using a programmatic command or let it watch your brain for signs of distress then stop by itself. And, that’s what scientists at MIT’s robotics research lab have just invented – a system that’s connected to your brain which tells your robot how to carry out tasks.

How it Works

The original system is quite simple. It features a scalp EEG and EMG system wired to a Baxter work robot, which allows for a human wave or gesture when your robot’s doing something it ought not to be doing. For instance, you may have your robot tasked with doing some regular job like drilling holes, however, when it reaches an unaccustomed setting, you can gesture at the task that the robot should carry out. This lets you communicate with your robot more like you’d with another person.

Since the system makes use of hints like gestures and human emotional reactions, robots can be trained to interact with persons with disabilities and even avert accidents by pressing an alarm before the information is communicated vocally. This will help you stop the robot from damaging something and even help your robot recognize minor changes to its tasks before it carries them out.

It’s the Robot that Adapts

In their tests, the researchers ‘trained’ Baxter to drill holes in an aircraft fuselage. The task changed from time to time and, a nearby human could gesture to the robot to let it change position before drilling – basically training the robot to carry out new tasks amidst its present task. Furthermore, there wasn’t actual programming involved on the human’s part – just a gesture to the robot to shift the drill right or left along the fuselage. And, most importantly, you don’t have to think in a prescribed manner or train yourself on how to interact with the machine. Actually, it’s the robot that adapts to you, and not the other way around. The team plans to present these findings at this year’s Robotics Science and Systems Conference, scheduled for June 26 – 30.