The folks at OCZ have been talking for a while about a device known as the Neural Impulse Actuator, which employs several sensors embedded in a headband to read certain electrical impulses from the wearer, theoretically acting as a PC input device. Our own Cyril Kowaliski even got to try out the device at Computex last year, affording us a priceless opportunity to publish a picture of him wearing a headband.
Last week at CES, I got an update on OCZ's progress with the device, courtesy of Dr. Michael Schuette, OCZ's VP of Technology Development. Dr. Schuette has teamed up with the inventor of the device to improve it and prepare it for release as a product.
The concept behind the device is, in its way, relatively simple. The headband has a trio of sensors across the front, and those sensors read electrical impulses in various frequency ranges, which equate to a number of different facial movements and—so it would seem, at least—patterns of thought. Dr. Schuette said the original interface for the headband was a simple RS-232 serial connection, and the software reading the inputs was single-threaded. OCZ has now converted the hardware to a USB connection and has developed a new, multithreaded software layer that uses DirectX to process inputs. The difference, Schuette claimed, is substantial. The DirectX input mechanism is much quicker, and the software can use multicore CPUs to handle the mathematical tasks like fast Fourier transforms needed to interpret the signals from the headband's sensors.
He showed us the control panel for the device and walked us through the calibration process required for each individual wearer. The headband differentiates between a number of different actions, including the contraction of facial muscles (like jaw clenching and squinting), side to side eye movements, and alpha and beta brain wave fluctuations. OCZ's software shows a real-time readout for each variable tracked, and one can see the values fluctuate in apparently meaningful ways as the wearer talks, moves, or responds to me making fun of how he looks wearing that funny headband.
One may then map the various variables types to traditional PC input commands, including keystrokes, via OCZ's control panel.
Dr. Schuette showed us his setup for Unreal Tournament 3, in which he used the headband to control the basic movement and action controls: forward, back, left, right, jump and shoot. The software offers extensive control over how these inputs are registered, complete with input delays of specified time intervals to prevent one from, say, inadvertently registering a whole slew of jump commands in rapid succession. Schuette had developed a custom control profile for UT3 that worked well for him, and he noted that such profiles can be saved as a file and transported to another PC or shared with other users. Using his profile, Schuette was able to play a passable game of UT3. He still used the mouse as most players do in an FPS, for six-degrees-of-freedom camera/look control, but the headband did the rest. His control over motion obviously wasn't perfect, and deathmatches do afford some room for imprecise movement, so long as one keeps moving. Still, the effect was very sci-fi. Schuette mapped the most basic controls to the facial muscle inputs, but the most impressive input had to be his "jump" key, which he mapped to one of the brain wave readings and activated via what he called his "Tourette's impulse." Basically, he'd think of an expletive, and his character would jump.
Having seen all of this, I wanted to try the thing out for myself, and the folks from OCZ obliged me. After the calibration process, during which I had to relax in order to establish a calm baseline, I started out with a simple game of "pong" built into the OCZ software. I was able to control the paddle's up and down motion essentially by squinting—more is down, less is up. With that complete, Schuette promptly fired up UT3 and threw me to the wolves using one of his most complex control profiles. I struggled quite a bit a first, but was able to control forward/back motion and shooting before too long. Left/right movement was a struggle, but I still got a few kills.
One of the things that surprised me about using the headband is how, well, cerebral it is to use. I expected to be contorting my facial muscles wildly in order to use it, but in reality, it's best to relax and gently control things. On top of that, something about the experience of using it is difficult to articulate but intuitive to apprehend.
In other words: Freaky! It's a brain control thingy!
We are still clearly in the very early stages of this technology's development, if it is to become a broadly useful form of input and control. Even with an experienced user like Dr. Schuette, the directional controls appeared to work in a fairly rudimentary fashion. As a result, I'm skeptical about this device's immediate utility as a gaming controller. The one big advantage it seems to have is reading one's reflexive responses to surprises. Dr. Schuette claimed he could map such a response to the "fire" button, for instance, and wind up shooting someone who had just popped around a corner before he fully consciously registered what he was seeing. Such quick-response opportunities could be happily lethal in a deathmatch.
Beyond gaming, though, the Neural Impulse Actuator could hold quite a bit of promise as an input device for those with physical impairments like Parkinson's disease. Dr. Schuette said the device's creator is already working with a few individuals with such impairments, and the initial results are promising. Schuette believes we're just reaching the point where consumer PC hardware offers the computational horsepower to make devices like the NIA feasible. He said quad-core processors are best used with games like UT3, and interestingly, he claimed he'd seen noticeably better responsiveness from AMD's quad-core CPU's than Intel's, perhaps as a result of AMD's monolithic quad-core architecture.
If I recall correctly, Schuette said OCZ hopes to bring the first version of the NIA to market in the next three to four months for between $300 and $400. We will be waiting impatiently for a chance to review the NIA once it's ready to roll.