I like to tell people “If it requires direct human interaction is is a Home Automation failure.” It is an admittedly hyperbolic an unrealistic statement but it always generates great conversations.
In short several sub-goals are intended to be captured in this statement:
The house should just do it on its own based in what the house knows at the time. For example, lights should turn on when it’s dark and someone enters a room. HVAC should maintain a desired temperature when someone is home.
There are clearly situations where complete automation is not possible. In those cases the computer controlled interface must be as easy or easier to use than the traditional interface. For example, turning on the light should be as easy or easier than flipping a wall switch.
Guests, children, and house members who currently can’t find their phone must be able to control those aspects of the system that do require human interaction. I find that the traditional interfaces (e.g. wall switches, the physical Nest interface, etc. work the best but I can see a dedicated mounted tablet or a remote control being suitable as well.
There are also clearly situations where you will want to override the automated behavior and the way one overrides it must likewise be as easy to use as the traditional interface.
If it is not fully automated or the user interface is not easy to use, the system must be able to do something that is impossible/impractical to do through manual means.
If the automation cannot meet these criteria then IMHO it should not be considered. So these so called “smart light bulbs” are a non-starter to me because they require an interface that is less convenient than switching a light switch. One could argue the ability to control color makes it fit into my last category above in which case that is a compelling use case for you, but I personally have no interest in adjusting the color of my lights so it doesn’t work for me.
So whenever I want to start a new Home Automation project I make sure it meets those criteria and if they don’t and I can’t see a way to eventually grow them to that point then I do not do the project. I find this results in a more usable home with a huge WAF.
Sometime these factors also result in a case where the automation does not even need to be integrated with with OH.
To give an example of what I’m talking about I’ll reference my lighting. I have four lights that are automated in some way: garage, porch, and two lamps.
The garage is automated with a simple motion sensor and not integrated with OH at all. All I need is for the light to turn on when I enter the garage and stay on long enough for me to get what I need so a cheap motion sensor light switch is perfect. I don’t really need to know in OH whether the light is on or not because I know it will shut itself off on its own. And I have an IP camera there which I could have report the light turning on or off if I really needed that.
The porch is controlled with a zwave switch and the lamps are controlled with zwave outlets. They are programmed to come on and turn off based on sunrise and sunset. One lamp turns on 30 minutes before sunrise and off at sunrise and the rest come on 90 minutes before sunset and off at 11. The two lamps also turn on when the weather code says it is likely to be cloudy right now but only between sunrise and 90 minutes before sunrise.
This last behavior can be overridden on a light by light basis by manually switching the physical light switch or on a light by light basis or all lights at once through the sitemap. It is a tad awkward for the lamps because you have to press a button on the outlet for the lamps instead of the lamps physical switch, but I’ve found it very rare that the behavior has to be overridden and it is still less awkward than requiring the use of a phone.
Here is another example to show how I deal with things that cannot be completely automated.
I have two garage door openers, one of which the remotes quit working and it is too old to get a replacement. So I have a relay wired to the door buttons controlled by a Raspberry Pi. So it is possible to open or close through the sitemap. This is clearly less convenient than pressing a traditional garage door remote. Therefore I had to take a two pronged approach to make it as easy: Android and iPhone.
For Android I use Tasker with AutoLocation which is able to detect when I am driving and to pop up a dialog when I’m driving and ask if I want to open the garage. Since I drive with my phone mounted all I have to do is press the one button which appears to open the garage which is as easy as pressing the one.
For iPhone this really isn’t possible so for my wife I got a zwave scene controller (Minimote) and all she needs to do is press the one button to open/close the garage.
The button on the wall still works as well. So now I have the garage computer controlled with all of the new capabilities that enables yet it is still as easy to use as a traditional garage.