Timed measurement/Scheduling

[cfreeman]

Spaces with MtF sensors are typically not going to be 24/7 spaces, and we need a way to ensure that we don't have tons of dead data/useless DB entries from when the spaces are closed.

[cfreeman]

Couple of different options here:

• We can leave the scout running and have the mothership filter (drop) data when the MtF sensors report data outside the specified interval.
• Start and stop the MtF sensors at the specified interval using the existing API.

Not sure exactly which approach will be the best (which will be the most robust with computer vision).

The good news is that if the computer vision in the sensor is working correctly and we don't get any interaction data when the space is empty. We should only get health logs from the sensor around the clock. The sensor sends these health logs to the database every 15 minutes. These records occupy about ~28 Bytes each. Which means health logs will consume about 1MB a year.

[griffey]

Once the CV is tuned properly, I agree....the interaction data SHOULD be non-existent when the area is empty. But we need to be able to deal more robustly with changing light levels and such before that's actually true. I do like having the health data continually.