Skinput: Approapriating the Body as an Input Surface
I. Introducción
Devices with significant computational power and capabilities can now be easily carried on our bodies. However, their small size typically leads to limited interaction space and consequently diminishes their usability and functionality.
Appropriating the human body as an input device is appealing not only because we have roughly two square meters of external surface area, but also because much of it is easily accessible by our hands, also it can be arms, upper legs, torso etc…
Well in this paper I’ll describe the wearable sensor for bio-acoustic signal acquistion. And also to describe an analysis approach that enables our system to resolve the location of finger taps on the body. Finally I’ll describe the limitations of this sensor.
II. Related work
Always-Available Input
The primary goal of Skinput is to provide an always available mobile input system, that is, an input system that does not require a user to carry or pick up a device.
The SixthSense project [19] proposes a mobile, always- available input/output capability by combining projected information with a color-marker-based vision tracking sys- tem.
Always-Available Input
Signals traditionally used for diagnostic medicine, such as heart rate and skin resistance, have been appropriated for assessing a user’s emotional state.
These features are generally subconsciously- driven and cannot be controlled with sufficient precision for direct input. Similarly, brain sensing technologies such as electroencephalography (EEG) and functional near-infrared spectroscopy (fNIR) have been used by HCI researchers to assess cognitive and emotional state
There has been less work relating to the intersection of fin- ger input and biological signals. Researchers have har- nessed the electrical signals generated by muscle activation during normal hand movement through electromyography.
III. Skinput
Bio-Acoustics
When a finger taps the skin, several distinct forms of acous- tic energy are produced. Some energy is radiated into the air as sound waves; this energy is not captured by the Skin- put system.
Bones are held together by ligaments, and joints often include addi- tional biological structures such as fluid cavities. This makes joints behave as acoustic filters. In some cases, these may simply dampen acoustics; in other cases, these will selectively attenuate specific frequencies, creating location- specific acoustic signatures.
The prototype features two arrays of five sensing elements, incorporated into an arm- band form factor.
Base don pilot data Collection, the owners selected a diferent set of resonante frequencies for each sensor Packaged.
Each channel was sampled at 5.5kHz, a sampling rate that would be considered too low for speech or environmental audio, but was able to represent the relevant spectrum of frequencies transmitted through the arm.
For example, the ATmega168 processor employed by the Arduino platform can sample analog readings at 77kHz with no loss of precision, and could therefore provide the full sampling power required for Skinput (55kHz total).
Data was then sent from our thin client over a local socket to our primary application, written in Java. This program performed three key functions. First, it provided a live visu- alization of the data from our ten sensors, which was useful in identifying acoustic features. Second, it seg- mented inputs from the data stream into independent in- stances (taps). Third, it classified these input instances. The following figure show us a sample of the sensors.
IV. EXPERIMENT TEST
Methodology
To evaluate the performance of this system, 13 participants were recruited from the Greater Seattle area. These participants represented a diverse cross section of potencial ages and body types, ages between 20 to 56.
This methodology was divided in 5 diferents locations:
But just five fingers, whole arm and forearm were part of the test. Participants were seated in a conventional office chair, in front of a desktop computer that presented stimuli. For con- ditions with sensors below the elbow, the arm- band was placed ~3cm away from the elbow, with one sensor package near the radius and the other near the ulna. For conditions with the sensors above the elbow, the armband was placed ~7cm above the elbow, such that one sensor package rested on the biceps. Some of the results of the evaluation were:
Five Fingers
Despite multiple joint crossings and ~40cm of separation between the input targets and sensors, classification accura- cy remained high for the five-finger condition, averaging 87.7% (SD=10.0%, chance=20%) across participants. Seg- mentation, as in other conditions, was essentially perfect.
Whole Arm
Participants performed three conditions with the whole-arm location configuration. The below-elbow placement per- formed the best, posting a 95.5% (SD=5.1%, chance=20%) average accuracy. This is not surprising, as this condition placed the sensors closer to the input targets than the other conditions. Moving the sensor above the elbow reduced accuracy to 88.3% (SD=7.8%, chance=20%), a drop of 7.2%. This is almost certainly related to the acoustic loss at the elbow joint and the additional 10cm of distance between the sensor and input targets. Figure 8 shows these results.
The eyes-free input condition yielded lower accuracies than other conditions, averaging 85.0% (SD=9.4%, chance=20%). This represents a 10.5% drop from its vision- assisted, but otherwise identical counterpart condition.
Forearm
Classification accuracy for the ten-location forearm condi- tion stood at 81.5% (SD=10.5%, chance=10%), a surpri- singly strong result for an input set we devised to push our system’s sensing limit (K=0.72, considered very strong).
Walking and Jogging
The participants also do this test walking and Jogging, this can produce false positives (i.e., the system believed there was input when in fact there was not) and by the other hand true positives (i.e., the system was able to correctly segment an in- tended input).
Walking trials, the system never produced a false- positive input. Meanwhile, true positive accuracy was 100%. Classification accuracy for the inputs (e.g., a wrist
tap was recognized as a wrist tap) was 100% for the male and 86.7% for the female (chance=33%).
In the jogging trials, the system had four false-positive in- put events (two per participant) over six minutes of conti- nuous jogging. True-positive accuracy, as with walking, was 100%. Considering that jogging is perhaps the hardest input filtering and segmentation test, we view this result as extremely positive. Classification accuracy, however, de- creased to 83.3% and 60.0% for the male and female partic- ipants respectively (chance=33%).
Les recomiendo ver el ——>VIDEO SKINPUT MICROSOFT<——
Si quieren saber más de los autores de este proyecto da click AQUÍ, para descargar un documento Word donde explico brevemente otros proyectos de ellos.
Espero les haya interesado este artículo.
