[Feedback]   [Home]   [Video Page]

DIFFERENTIAL CAPACITIVE TOUCH POINTER
The differential capacitive touch pointer comprises just four paper-thin copper pads and a small IC. The IC measures the capacitive effect of a finger (or thumb) touching the pads. Any touch affects all four pads but the touch position determines the strength of the effect. The differential between left and right and top and bottom pads yields a resolution of at least 1000 in both dimensions. Because, it is differential, the position signal is immune to electric noise and background capacitance. This allows the sensitivity to be very high without swamping the signal with noise. Consequently, the device can be placed inside a thick plastic case with no opening for the touch pads. The touch position is detected through the solid case.

DISPLACEMENT ACCELERATION AND SUB-PERCEPTIVE GESTURE ANALYSIS
The high resolution is useful for analysis but no human has sufficient tactile acuity to directly control 1000-point resolution over an 11mm range. Therefore, the input can’t be mapped directly to the display but must be used in displacement mode. All such devices change the mapping according to sliding speed. Sliding faster expands the mapping to produce a larger displacement. With such a small input area, mapping acceleration methods used for larger devices are not very effective. I developed a new acceleration method based on control theory. This is instantaneously responsive without overreacting and is continuously variable without the quantization seen in other methods. The control terms include not only the usual, such as speed integral and derivative, but also unique measures of user intent derived from sub-perceptive gesture analysis.

FLICK SIGNAL PROCESSING: CROWD WISDOM
Also see flick algorithm development. As with all capacitive touch devices, the finger’s X-Y position as it approaches or withdraws from the device is not reliably indicated. Most touch devices discard this ambiguous hover information. However, with a small device like this, a natural and comfortable flick may comprise only approach and withdrawal with little or no precise position information. To recognize the user’s intent as the length, speed, and angle of a flick, I developed 25 unique signal processing methods. These individually vary in reliability from 50% to 80%. Since no one method is good enough, I developed an adaptive voting scheme. If the best ones agree, they are always right. Otherwise, more and more of the less reliable ones are allowed to vote. These less reliable methods are not simply inferior but ones selected for their reliability in resolving disputes between the better methods. The result is nearly 100% reliability, which is to say that most users think that the device has responded correctly. There are no cases where the device reports incorrectly but a few where it cannot be sure and reports this instead of an erroneous flick. See Flick Processor.