Why activity tracking is important? If you have been setting up automation systems for home automation or in professional environments, you must have discovered that having as much as information about the context of a specific situation is the key to be able to react appropriately. Well, I would even go further and say that a great piece of information we often miss, is the type of activities the people involved are doing.
Examples of activity tracking
If you are automating a home automation system and you would like to handle the lights for the users. In that case you must make the difference between “The user is reading” and “the user is sleeping”. If you are automating a door interaction, you must know if “the user is entering”, “the user is leaving” or “a delivery is coming”. If it cannot assess properly the situation the system will end up doing nothing or taking a random action that will for sure not satisfy your user.
I am thinking about many different ways to get partial information that can certainly be used to determine accurately what a user is doing:
- Use connected smart wristbands or watches, to gather pulse, arm movements to know what is the current state of the user (Sleeping, active, very active…)
- Use smart bed companions and sleep trackers in order to track bed activities (Sleeping, reading, not in bed…)
- Use a smart sound tracker [Project] to use sound fingerprinting to define the current situation (Watching TV, eating, playing, sleeping, having a phone conversation…)
- Use presence and movement sensors in the good old fashion
- Use phone status, GPS and accelerator’s data to help define the current activities (Watching the screen, on the way home, driving to school, walking back, in a call…)
- Use cameras with image processing software to check users’ location and activities (Eating, watching TV, sleeping, fighting ;)…)
- Use Bluetooth and WiFi tracking to define who’s is actually present according to their phone Bluetooth or WiFi signal
- Use Connected shoes [Idea] to determine is the owner is (Walking, running, exercising, sitting….)
Of course, each of these solution is bringing a small piece of information, but I am convinced that information aggregation mixed with pattern recognition and proper use of historical data (and deep learning) could really provide usable data that could be used for automation.
Let me know if there are more devices or information source that you think could be useful.