As mentioned elsewhere, a sensor is any mechanism for detecting and recording a signal from a human (in our context). Below each key on a computer keyboard, for example, is a simple kind of click sensor which registers a signal to produce a particular character on the screen when depressed. Similarly, when using EMG, or electromyographic (i.e., muscle voltage) sensors, the sensor being attached to the skin covering a particular muscle group registers a signal coming from electrical energy discharged by the flexed muscle. Again, with photoresistor cells, a special light sensor sustained (by a mount, see section II) mechanically at a distance from some body surface, a cheekbone for example, registers a signal every time an otherwise complete light field is distorted by a cheek movement. The photocell then senses negative values. In our work with Eyal Sherman, as well as with those disabled kids before him, we used the EMG sensor extensively [photos]. It attached easily to the facial parts of the people we were working with; it was inexpensive; and it could be placed practically anywhere. We began working with the photocell, however, because the EMG sensors could get messy; their performance decays with perspiration; they take some know how to attach and calibrate; and looking forward, we wanted to move away from having to attach anything to the face itself so photoresistors begin to resolve some of these issues. The photocells thus create a need for the next area of interface development: the mounting devices.
Here is where the Industrial Design members of our team began to really move in on the issue of interfacing for both Eyal and Brooke. If we were going to use photosensors, we needed to devise a way to actually suspend them over the areas of the face Eyal and others like him were going to be moving. Eyal's eye glasses were the first clue for the design group. After several iterations of the basic idea of suspending photocells from the frames of Eyal's glasses, we are very close to a plug and play prototype [photos] With Brooke, the work was different. Brooke has movement in here limbs, yet it is very restricted and turgid in many ways. Getting her movements in a position to activate sensors was another challenge altogether. On this project a second industrial design team would improvise a rich array of possible mounting systems. At the close of the semester they developed several working prototypes which worked. A second semester of trial and revision will greatly close the gap on a plug and play prototype for Brooke as well.
The next element in the interface chain is an electronics processor that takes the raw signals from the sensors and transduces them so as to be channelable through either the serial, parallell or SCSI port of the computer (which for us is usually a PC). Our THING devices achieve this end. THING I is a EMG signal transducer in that it registers voltage shifts. THING II and III on the other hand are signal transducers that register resistance shifts. While THING III is a super duper version of II, it is still in development at the moment making THING II Cyberarium's current signal transduction workhorse.
Finally, there is the program which makes all of the prior elements actually usable towards some end: NeatTools (Neat). Neat began as a DOS gesture recognition package. It took signals from sensors and could translate them into commands within the computer. Also, it allowed for a "tightening" of the signal-command gap: callibration. Neat, as we are developing it now, is turning into a world class program capable, essentially, of taking any sensor derived signal and yielding any command: turning any input into any output. Click for more.