You’ve may have heard of virtual keyboards controlled by thought, brain-powered wheelchairs, and neuro-prosthetic limbs. But powering these machines can be, well, tiring when it comes down to long-term use, a fact that prevents the technology from becoming widespread and being useful to people with disabilities, among others. But Professor Jose Millan and his team at the Ecole Polytechnique Fédérale de Lausanne (EPFL) in Switzerland have a solution: train the system to allow for multitasking.
One of the major difficulties in using brain-computer interface (BCI) is fatigue. The user can send one of three commands – left, right, or a no-command. After an hour, the concentration demands are just too exhausting for most test subjects. The key is to find a way to allow users to not send unnecessary commands to the interface when they are doing a secondary task. Millan’s innovation allows a user to multitask, to be able to do something else or even nothing at all for brief periods of time – resulting in longer sessions that aren’t as taxing.
In an ongoing study demonstrated by Millan and doctoral student Michele Tavella at the AAAS 2011 Annual Meeting in Washington, D.C., the scientists have hooked up 10 volunteers to the BCI and while doing “secondary” tasks such as reading, speaking, and reading aloud asked them to first deliver no-commands to the machine, then later, to deliver many left/right commands as possible. The trick is to decode the signals coming from EEG readings, which are recorded on the scalp and represent the activity of millions of neurons and have notoriously low resonance. By using statistical analysis to classify different tasks, the interface learns when the subject is not sending a command to the interface and can improve over time. The demonstration will show how this new system allows for both targeted control, around an obstacle to a precise goal for example, and makes it easier to give simple, once-given commands such as simply going straight down a corridor.
This cutting-edge technology will not be making the move from the lab to the production line just yet, but Millan’s prototypes are the first working models of their kind to use probability theory to make BCIs easier to use over longer periods of time through. Perhaps down the road the technology will allow someone with a disability to reread the first part of her email without sending false commands to the virtual keyboard, making the daunting task that much easier. The next step will be to interpret more cognitive data, allowing for the interface to recognize errors in command and respond appropriately.
Michael Mitchell
[email protected]
+41 21 693 70 22
