A telepresence robot developed at the École Polytechnique Fédérale de Lausanne (EPFL) that can be controlled by thought may give people with severe motor disabilities a greater level of independence. Successfully put through its paces by 19 people scattered around Central Europe – nine of whom are quadriplegic and all of whom were hooked up to a brain-machine interface – the robot handled obstacle detection and avoidance on its own while the person controlling it gave general navigation instructions.
The robot is based on Festo's Robotino platform, which features three wheels, nine infrared sensors and a webcam that can be used for obstacle detection. To this, the researchers added a laptop with an integrated camera so that it could serve as a telepresence robot. The person controlling the robot sees through this laptop camera and can meet and interact with people via Skype, which displays their face on the screen.
The semi-autonomous robot is controlled through an electrode-studded hat that was designed as part of the TOBI (Tools for Brain-Computer Interaction) project, which ran from 2008 to 2013 and focused on developing practical technology to improve the quality of life and rehabilitation processes of disabled people. Subjects learn to manipulate it by first imagining hand and feet movements to help calibrate the system to their brain waves, then by rehearsing two motor imagery tasks.
Even if the person in control of the robot doesn't tell it to, the robot can autonomously avoid obstacles. It will also continue on its indicated path if the user doesn't provide any additional instructions, until given the instruction to stop. This can help prevent the user getting tired, as it allows them to take a break as the robot continues on its merry way.
The EPFL researchers recruited nine subjects with motor disabilities and 10 healthy people across Italy, Germany and Switzerland, then taught them to pilot the robot – which remained in an EPFL laboratory in Switzerland that they'd never visited – over the internet using just their thoughts. It took fewer than 10 days of training for all of the subjects to master remote control of the robot, and the tests revealed no difference in piloting ability between healthy and disabled subjects.
The researchers hope that mind-controlled robots – be it this one or others that use different interfaces or that incorporate greater levels of artificial intelligence – might become a part of daily life for people with severe motor disabilities, allowing them to move around and talk to people even if their body is confined to the bed. But lead researcher José del R. Millán isn't making any predictions. "For this to happen," he notes, "insurance companies will have to help finance these technologies."
A paper describing the study was published in the journal Proceedings of the IEEE.
You can watch a video that showcases the research below.
Source: EPFL