Home control vs home automation.

[waynehead99]

That would be great and it will get to that point probably within the next decade. But I guess that brings up the question since everything has to have some sort of programming weather it be by you or someone at company X, what is everyone's definition of automation? I think a light turning on or off based on conditions vs a switch is automation. It's not adaption but it is automation.

[Mattyjee]

Probably easier (and cheaper) to just hire a full time butler to turn lights on and off for me.

[stevea]

CBL (can I call you that?):

That is a really interesting bifurcation of what we all do here. Some folks who have read my posts know that I've been pretty disenchanted with Homeseer, especially in the transition to HS3. I've stayed with HS2, and mostly think I'm done with any future upgrades. I think some think I'm just being a stick in the mud. Until I saw your post, I don't think I could adequately describe my feelings.

But now...

At least part of my issue may be that I have managed to transition some from the "Control" to "Automation". I think my vision of perfection in my config of HS would be if - eventually - no one else in the house really notices it is there. Lights turn on for you when you enter a room. Someone new to the house may notice it at first, but shortly it just seems to be the way things are. The thermostat changes set points at different times of the day. The garage door opens when one of our cars approaches. The whole house fan turns on in the summer and the A/C turns off when the temperature difference between outside and inside makes it more economical. The sprinklers run differently when it has recently rained.

Perhaps the most telling evidence: My wife doesn't say anything until something that she expected to happen (i.e., control) didn't.

I don't think that maintaining this "control" should take hours and hours of my time every week to maintain. This made HS2 a difficult but satisfactory solution. The years I spent honing this to work just they way it should made HS3 a non-starter: Virtually none of the hundreds (heck, maybe thousands) of scripts and events that I had would transport over.

I had been thinking it was all because I had moved from being a nerd / experimenter, to someone who was actually a "customer": I wanted it to work, but I wasn't interested in dedicating hours per day to keep it working. However, I think maybe I have also transitioned to someone who is more interested in "Automation" than "Control".

Anyway, thanks for laying out this line of thought. I think it helps me (and maybe others) understand. In my case, it helps me bound my thinking. Maybe it will help the Homeseer Tech people see why I (and others) are so frustrated.

Steve

[collegeboyslive]

Steve,

Trust me , I too was just like you, I bought hs3 as soon as it came out but only last week did I feel that it had matured enough and had enough support for me to invest the time and effort to make everything over. I am happy to report its going a lot smoother than expected.

And yep I gauge my programming and ideas by how natural and easy no technical people use it or not notice what's happening , just that it is.

[NeverDie]

Answering my own question from above, consider the following simple scenario:
Every time you open the entrance door and it's nighttime, you turn on the entrance light.

Of course, you could program that as an event yourself, but there's nothing new about that. Suppose instead that an add-on program notices that you do that "behavior" consistently, 100% of the time, for, say, 30 iterations (or 5 or 100 or whatever) in a row. Suppose the next time you open the entrance door at night, you're pleased to find that the light gets turned on for you automatically, and you never added that event logic. It just happens.

Would you all want that? When somebody says, "I want it to learn my behaviors," we eventually need to agree on what that means. At the moment, this is the simplest example I can think of.

OK, I thought of an even simpler scenario: it's X follows Y, always. So, instead of the front entrance, it's the basement entrance. Every time you open the basement door, you turn on the basement light. Simple.

So, that's one extreme. Maybe opposite extreme is where you expect it to be smarter than you are, where it notices your complex repetitive behaviors before even you do. e.g. if you happen to see the outside lights on during the daytime (maybe someone bumped them by accident into the ON position), and you turn them off. Imagine it trying to learn that, though, from your behavior. It sees that the lights are on during the daytime, so how does it know that you didn't intend that? In the worst case, it might start turning on the outside lights during daytime as it starts to replicate your mistakes, and then turning them off before you ordinarily notice it. Then you aren't even aware of what a screwed up behavior it has learned from you.

[Raptor]

i have felt this way for a while,

my system is mostly home control with some aspects of automation, namley geofencing and sunrise/sunset control.

this is mainly because i have only just for the base stuff stable and working the way i like before having it take over functions.

the amount of work to "tell it" to automate my life is also high.

[Uncle Michael]

This is a good topic, and some revealing discussion.

Just my $0.02, but I think the distinction between automation and control has nothing to do with learning. Automation is a coined term for automatic control and came into widespread use in the 1940s.

The distinction I make between home automation and home control is at the automatic part. The simplest example I can think of is the motion control light switch. The switch senses motion with its PIR and turns on the light. If it doesn't sense motion for a reprogrammed period of time, it turns off the light. Yes, the device is just controlling the light, but it's doing so automatically, without the need to physically reach for and flip the switch.

The stuff that is being sold with apps for a phone are controllers if all they do is provide an alternative switch location. In the worst case, instead of the convenient wall switch, always available and ready for turning the light on or off, you get an app for a device in your pocket that you first have to fish out, then load the app, then navigate to the right icon and press it. Had that come first, the wall switch would be hailed as a major improvement in convenience!

Adaptive automation, that learns routines by monitoring activity, still requires some pretty heavy duty data acquisition and analysis capability. AFAIK, all systems that do that well rely on sending the data to a central processing location, with the attendant loss of control over the data and the risk of unwanted use of the data for sales and marketing (at best) or even surveillance purposes.

[S-F]

Adaptive automation, that learns routines by monitoring activity, still requires some pretty heavy duty data acquisition and analysis capability. AFAIK, all systems that do that well rely on sending the data to a central processing location, with the attendant loss of control over the data and the risk of unwanted use of the data for sales and marketing (at best) or even surveillance purposes.

Just because this is currently how that works doesn't mean that it need to or will in the future. I want a personal Watson talking to HS and I want to run it on my hardware. I want my HA system to have an array of sensors that currently don't exist so I can have bulletproof occupancy detection. And to top it off I want to have the system remember EVERYTHING so I can query it later or it can use the data to make new decisions. The virtual personal assistant is still in its infancy. Google Now is cute and all but it doesn't really do too much. I'm assuming that Siri and Cortana are similar. I have seen secretaries for CEO's and the like who are astounding in what they can accomplish. I want that and more. You can't expect your secretary to adjust your thermostat or turn the lights on and off. We all know it. This is the direction we're heading in. I love HomeSeer, have great faith in the staff at HST and can't speak highly enough about it to others who are interested in HA. But if someone comes out with a HA solution that offers the aforementioned qualities I'll be on it like white on rice.

[Uncle Michael]

. . .I want a personal Watson talking to HS and I want to run it on my hardware.

Me too! I also want a guaranteed income of at least a million dollars a year for life, and good health and sound mental capacity lasting forever, or at least until I'm 110.

Yes, I'm sure the technology for adaptive automation is coming, and at some point will be available to individuals for use in private residences, but it's not ready just yet. As you note, "The virtual personal assistant is still in its infancy." And that's running on the most powerful computers Google and Apple and Microsoft have chosen to apply to the problem.

[S-F]

Me too! I also want a guaranteed income of at least a million dollars a year for life, and good health and sound mental capacity lasting forever, or at least until I'm 110.

Yes, I'm sure the technology for adaptive automation is coming, and at some point will be available to individuals for use in private residences, but it's not ready just yet. As you note, "The virtual personal assistant is still in its infancy." And that's running on the most powerful computers Google and Apple and Microsoft have chosen to apply to the problem.

I dunno brother. When Watson won Jeopardy it was a stand alone system not connected to the internet. So the true virtual assistant may come well before your 110th birthday. It's true that all the Google, Apple, etc. virtual assistants are run on super whiz bang systems but they are serving tens or hundreds of thousands of people. Imagine what could be accomplished with a reasonably powered server at home. I'm sure the grossly over powered system I have HS3 running on ATM cold do it.

[S-F]

Here's an interesting article from a week or so ago about a HA company I'm not familiar with that is moving into data analytics create the "smart home" as opposed to the controlled home. It should be noted that this is a cloud service, which is how they have acquired enough data to have meaningful analysis.


http://www.lightreading.com/iot/iot-.../d/d-id/716686

[Uncle Michael]

When Watson won Jeopardy it was a stand alone system not connected to the internet.

Yes, but it was a bit bulky, filling a whole room with hardware. And answering Jeopardy! questions (guessing Jeopardy! questions?) is a much more circumscribed problem than predicting the behavior of any arbitrary human.

Nonetheless, I agree that the hardware is unlikely to be the limiting factor long-term. It will be the cost of developing the logic and software to identify and deal with the vague 'questions' that are at the heart of predicting what to do next. I'm sure it can be done, but finding a way to pay for it will be the challenge.

I think that's why Watson is being focused on answering medical questions now. Identifying customers willing to pay for the staggering development cost is easier in medicine.

[S-F]

Ian Mercer is doing some pretty impressive stuff with his system and he's using a Nuc. Of course he's just one man and his HA system probably isn't going to be winning Jeopardy any time soon .

[NeverDie]

Ian Mercer is doing some pretty impressive stuff with his system and he's using a Nuc. Of course he's just one man and his HA system probably isn't going to be winning Jeopardy any time soon .

That's interesting. His blog says he has over 250 sensors in his house. Well, that might certainly help! I don't know how many or what kind, but you're gonna need a lot, and the more the better.

Which of his stuff did you find most impressive? Does he really have the world's smartest house?

[S-F]

I'm most impressed with his natural language interface and his "Project Remember Everything". Through a chat interface, or similar, he can instruct his system and query it. He can ask it things like "Who called on the second Thursday of last November?", or "How many miles did I run last week?". Or better yet, "Why did you turn the lights off just now?". He can ask it why it did something.