Skinput

Skinput: Appropriating the Body as an Input Surface [1]

Synopsis

Skinput is an input technology that uses the propagation of acoustic waves to resolve the location of taps on the arm and hand. This paper is an example of “bio-acoustic input”.

Strengths

  1. Technique capitalizes on the fact that the body surface represents a large interaction area that travels with us at all times. Proprioception allows accurate targeting of body locations without looking.
  2. Considers the possibility of a brain-computer interface, but notes that bandwidth is low and requires an inordinate amount of training and concentration.
  3. The paper includes a description of the translational and compression waves that propagate from an impact point, as well as the conductive patterns and effects of bones and joints, but the sensing technique does not depend on this understanding—just the complex signal patterns that result, which apparently are why the system works so well.
  4. Uses a Support Vector Machine (SVM) to classify the 186 features computed from the 10 sensors in the armband.
  5. Used low-tech stickers to mark forearm tap locations, modeling a possible projection system (needed because upper arm locations are not well defined/named, like wrist and hand areas).
  6. An unexpected result: They can distinguish certain types of taped surfaces. Also, different types of taps can also be detected (e.g., with fingernail, with knuckle, or with tip of finger). When paired with a touch-sensitive surface, this facility allows another source of input.
  7. An overall reaction: The work included a number of detailed experiments and the paper reported the setup and results in a way that would allow others to verify the experiments.

Weaknesses

  1. They move the sensor apparatus to the forearm to be able to more accurately determine taps on the finger tips. Problems occur due to the attenuation caused by intervening joints (between the tap location and sensor array).
  2. The authors note that there is a link between BMI and poor accuracy, consistent with their prediction that body structure would adversely affect the ability to pinpoint touches. They also note that men performed better than women, although they attributed this difference to the correlation with BMI. Were the women in their test sample the ones with high BMI?
  3. Their supplemental walking/jogging testing experiment is based on only two participants and very few trials, so the data does not seem reliable.

Do we have access to the videos mentioned in the paper?

Bibliography
1. Chris Harrison, Desney Tan, and Dan Morris, Skinput: Appropriating the Body as an Input Surface.