Instrumented People, Places and Things: bottle cams, 2012

Journey

The team's chart of one bottle: a row for each room and one for storage, shaded by day, the line colored by mode. A dot is a label, placed where the team saw the bottle used. Solid lines come from what a clip shows. The dashed lines are reconstruction: the bottle is drawn where the team last saw it, until a later clip shows it somewhere else. What that rests on is the next three lanes, on the same time line. Dots close together are spread apart so that each can be seen; zoom in for true time. The lanes are always at true time.

Motion

Each mark is a stretch of movement the team's software found in the log, taller for stronger. Dark marks pass the team's own test for use: over 4 seconds and a value over 75. Close in, it becomes the log itself, as the team's viewer drew it.

Camera

Each clip the camera took. Dark: the team found a use in it. Mid gray: it woke and caught no use. Outlined: the log says the camera ran, and no video file survives.

Sound

A spectrogram of each clip, worked out here from its sound: sixteen bands from 60 Hz to 12 kHz, eight frames a second, color for volume. Words cannot be read from it by eye, and nothing finer leaves the study's folder.

Squirts

The squirts the team counted by hand for each label.

A use

The team's viewer: the bottle turning to each reading, the readings around it in 3D going orange to blue through the clip, and the change in total magnitude along the time line. The bottle faces as the team's viewer drew it, and is white here where the team's was teal. The labels and the sound are added.

All four

Added here. Where each bottle rested, from the place the team coded at the start and end of each clip, with its uses over it.

Model

The team's 3D model of July 2012: time runs along an axis that stands for storage, a wheel of rooms turns around it, and each label is a squashed sphere in its room's direction, sized by squirts. Bottle 348 is the team's own geometry. The rule that made it was recovered from that geometry and rebuilds the other three.

Held

The team's drawing of one bottle in many positions: every log row during labels on horizontal surfaces, beside those on vertical ones, side by side and close as the team's film shows them. Each copy is turned by the team's viewer's rule, which draws the bottle as it was held: tipped toward its nozzle when it sprayed down. Which way it faced about the vertical is not in the log, since an accelerometer cannot sense it, so the crowds face the reader. White here, where the team's were green.

Labels

The team's parallel coordinates of every labeled moment, with the bottle cams' and the handheld cameras' together.

The study

The team's chart of strong moves per day, by instrument, in each home's own sequence of study days. The caddy's row counts its cups being emptied. In home H2 the two bottle cams stood outside the caddy, so its cups do not follow them.

Bottle cam

A camera with a microphone and an accelerometer, a clock and a memory card, in a white housing on the bottle's neck.

Accelerometer

Three axes, eight times a second. The logger writes a row when any axis changes by two counts, and otherwise one about every 10 seconds, so its log shows when a bottle lay still.

Waking

From the firmware and the logs: two seconds of steady change turn the camera on. It records about 8 seconds later, and about 45 seconds of stillness turn it off. So the first dozen seconds of any movement are in the log and not on the video, which is what piecing the streams together had to mend.

Labels

The team's tagger, used by hand: room, object, surface, its shade and orientation, tool, squirts and mode for each thing cleaned. A later pass, in January 2013, added where the bottle rested at the start and end of each clip, and the mess, what the person judged the result by, and how it met expectations.

Scales

Each bottle cam's bottle was weighed at the start and at the end.

Four bottles in three homes for about three and a half weeks. People knew the camera was there, and many clips have them talking aloud to it. The labels are one team's reading of the video, made by hand. In TJ's words: “Data does not arrive already labeled with the meaning of an activity.”

The cameras show what a bottle saw, not why it was picked up. A clip's end is not the end of a bottle's use: a bottle “kept out” at the end of a clip is sometimes put back minutes later, in a clip with no use in it.

“Storage” is itself a reading. Bottle 348 lived on a bathroom window sill. The team first counted that as its storage, and later filed it under “out”. This page keeps the first reading, and says so.

A later round, in 2013, planned a pump bottle with its own sensors and separate cameras triggered by the bottle's movement. Hardware and one page of notes from it were found here, and no data, so nothing on this page comes from it.

The clips, their sound and their frames show people in their homes, and stay on the study's computer until they have been reviewed. The handheld cameras' file names, the team's notes on the clips and the dates stay there too. The products are named only as Cleaner A, Cleaner B and their kinds, and the two bottle photographs on the slides carry no labels.