Is it possible to make a "smart home" truly smart?

Question

Automation

For many years there has been progressive progress in automation systems (18 years I have been only inside the issue), but there are no revolutionary and qualitative changes. The most important thing in automation remained unchanged: each individual system, be it heating, ITP, ventilation, air conditioning, power supply, underfloor heating, etc., was equipped with a self-sufficient control system. The trouble is that the conservatism of manufacturers, mounting and selling organizations, service providers offer the end user a closed system. With a good organization of design work, there is a minimum consistency in boundary issues between systems. But there are not many good situations. In addition, as experience shows, most systems work far from optimal values. And all this trouble entered a self-reproducing cycle. The consumer has long been aware that everything will work very poorly, but he deceives himself with the hope that a miracle will happen to him and harmony will settle in his home. And then a special "brownie" with the highest technical, preferably Soviet-style, and permanent teams and adjusters with lists of comments from organizations that install and sell equipment settle in the house. All this is about the houses of very wealthy people, in simpler housing everything is not so large, but the problems are similar.
Even a one-room apartment equipped with an optimal set, by modern standards, is a very ill-conceived complex in terms of management. Heating with thermal bulbs on control valves, air conditioning with its own remote control, exhaust systems, leakage protection, electric underfloor heating in bathrooms, supply ventilation (ventilation). One person can and will understand the features of the work of all systems and how to avoid mutual negative influence. But the rest of the tenants will simply be the human factor that will turn the symbiosis with the apartment into a constant headache.
Until now, manufacturers of various equipment have not offered a solution with a good concept of unification and structure in building interconnected systems. There is a community of KNX equipment manufacturers and a solution from LonWorks and many others, in the first two the idea was seen for something more than a unified protocol. But nothing has grown. The primary task of making money turned the idea into a closed club for crazy users who are ready to pay for the same systems, but many times more expensive. A central console and / or keys for a lighting scenario are added to the functionality.
For all the time, the task of the final user control panel was solved. And that's it!
Recentthe publication of a good specialist shows the dead end of the whole approach to automation. Something truly smart can only be done with the use of freely programmable controllers, because. the algorithm must take into account all the interactions of aggregates and systems with each other. In some units, a standard automation system with a network protocol. Technically, everything you need is available, but it is impossible to link all the algorithms and, most importantly, configure the system to perform the tasks in the best way. At such facilities, we temporarily install (for about a year) a computer with a SCADA system for data collection and analyze the operation of the combined systems with subsequent adjustment. But our method contradicts the realities of the market. This can only be done as a hobby with a good supply of time and money. Serially, such projects have low repeatability and, again, the cost of adjustment.
I won't push further. If there are questions, I can clarify separately how and where things are in detail. But there is one thought.
Let's put:

• All self-sufficient systems are equipped with a powerful controller with a widely available and developed software and hardware environment for development and support.
  
• In classical use, the system is mounted and configured using long-established methods. Already now it is possible to tie Apple Home into the central control system, but this upper level of control is slightly better than the capabilities of systems with KNX.
  
• When two or more systems appear, it becomes possible to breathe soul into the "smart home". We provide for machine learning of all systems, taking into account the entire set of states and required parameters. We add external factors: presence and active people, calendar, real time, etc.
  

Lower level
Each system has a set of input parameters and a set of external parameters from adjacent systems. For all variations, we obtain the output control parameters. If the system is excluded from the upper level, it will continue to work on the grid trained for local operation.
Medium level
When included in a complex of systems, work is carried out taking into account adjacent equipment and external factors. Such a network will need to be trained in advance and distributed among the systems.
All the value of the "smart home" is in a trained network for a special configuration of even the same houses. In the input parameters, your wishes and preferences. All permissible deviations (in reality, they lie within small limits) must be calculated immediately to exclude global actions with AI when controlling the user, for example, the temperature in the room.
We need the help of specialists in assessing the feasibility of such networks for a similar task. It is clear that the modernization of the complex in hardware (let's add, for example, a storage tank and start using excess energy between heating, ventilation, hot water systems, etc.) will require retraining of the entire complex for new capabilities and parameters.
Is it possible to build a distributed AI system, the loss of one subsystem from which will not require retraining of the entire complex.
The advantages of working all systems in the form of a single organism-complex will immediately lead to huge economic and environmental benefits. Next, work will begin on changing the generally accepted system configurations to the optimal one using in combination with AI and deep integration. Energy storage, redistribution, integration with renewable energy sources (water heating, energy recovery) with truly automated control will reduce resource consumption and harmful emissions.
The emergence of software implementation of neural networks gives hope for a reasonable implementation of the engineering systems management system by some analogy with the human body.

Answer 1

"smart home" in its principle is an idea-fix. because the task of the house is to provide comfort for a person, and a person is a fickle creature. to ensure my comfort, I need to be "measured". am i ready for this? not sure.
and of course the issue of maintenance the more complex the system, the more maintenance it will require. that is, either it will be necessary to provide a fully self-diagnosing system built on a modular principle, so that there would be no more problems from maintenance than problems that a smart home is designed to eliminate. and this is money.
but of course this does not mean, for example, that it is impossible to fully automate any processes,
for example, detect the presence of a person and turn off the light automatically or control the climate. it's just a question of price - it pays off - I bet it, it doesn't pay off - yes, well, it’s on ...
and at the expense of AI ... I would not trust modern AI to take out the trash bin. because if it messes up, then it’s very problematic to catch the AI ​​​​and give it a tambourine :)

Answer 2

the smart bomb could not be pushed out of the plane
on the topic, read about real-time modes and how the eyes of patients were burned out,
well, or with unmanned vehicles, ask where the hype
is without human control, all this will not be soon

Answer 3

for 18 years it was possible to understand that a smart home is implemented only by enthusiasts and only in developed countries, but not in the CIS. there will be few ready-made solutions, because Few manufacturers want to be responsible for stupid users by exposing control of their products to the outside.

Answer 4

Yes.
In short: this is the same (I take the simplest example) as writing a game of tic-tac-toe with a computer. But to give the computer only an understanding: he won this game or lost, but not to give the rules of the game !
Having played only a few games, he will write them down in his National Assembly himself.

Answer 5

You can do it, as you noticed, you need a more or less uniform connection interface and schemes for working with equipment. Gradually, either manufacturers will come to this, or one manufacturer will appear who will offer simple and understandable and not expensive work schemes.
Plug and Play - like on PC. type turned on and the system (main server) sees everyone, options for interacting with new equipment are automatically offered.

Answer 6

Since I have a problem with ventilation in my apartment of 48 squares, I master my climate system step by step.
Other than ITP, these are all functions of climate control and do not require any AI or machine learning. The climate control system simply uses heat like electricity. Consider the ideal.
We have a ventilation duct throughout the apartment, which distributes cold (air conditioner) and warm (from the radiator) air. There is also a constant flow of fresh air from the street.
What is needed.
One temperature sensor per room. Three control programs: no one is at home, someone is at home, night. And to calculate how much fresh air must be taken from the street, respectively pumped in the apartment - how many people in total, five each. For example, if you have 15 guests.
For the bathroom and toilet, it is more logical to lay a rubber mat, but you can also buy warm floors. There's a switch there.
How will you add AI here and what do you want to teach?
The problem with smart homes, in my opinion, is that for the poor, the systems are expensive both in terms of money and time. For the rich, smart home systems are insanely expensive in terms of the time spent on their installation (even if you order specialists) and subsequent maintenance. And it's all about electricity.
Also, a smart home begins with the words house and local area, then it generally becomes not clear what smart things can be made there for apartments for 99.99% of the inhabitants.

Answer 7

I’ll answer “unexpectedly”: because all the “developers” (in this case, they are exactly “engineers” and “bydlocoders”, without any jokes) look not at what is really needed and could become a truly “smart home”, but at OH, like AI. Well, what, to the devil, is the intelligence of the switch ?!
In short, instead of automating various things, they come up with some kind of ridiculous nonsense - "AI in a glass", which for some reason needs the Internet to turn the light on and off in the toilet and this is done "no, what are you, not at all to collect data ".

Answer 8

The question is interesting, but I completely disagree with the conclusions coming from this question. In my opinion, once again, there is an artificial imposition of expensive solutions from the manufacturer to the consumer.
You can just think, or ask a question, what do the majority need, not wealthy citizens, who are a small percentage of the total population, but the average consumer. Again, the question is, how many% of the population use Apple Home?
At the beginning of the article there is an example based on a one-room apartment. Then comes the conclusion:
Here, explain what personal benefits can arise for the average consumer living in an apartment building? What does he really need? Does it need learning, AI, and the costs of all this?
Will smart home with AI, in the near future. adequately learn, and correctly respond to the actions of let's say the child? Where is the guarantee that the smart home will not go crazy.
For me personally, a smart home should be predictable, adequately respond to my actions, and not guess and predict what I want to do. And one of the main requirements for me: there is a smart home - good, comfortable, convenient, no - you can safely do without it. As an example, the same TV with a remote control, the remote control is excellent, it got lost, it broke, the batteries ran out - he got up, went to the TV, did the necessary actions (decreased / increased the volume, changed the channel, turned it off ...).
So my answer to the question is: in some cases, a semblance of the mind is possible (if a person lives alone, he has established habits and all his actions can be predicted in advance), but for me personally this is not necessary.

Answer 9

A smart home is very expensive
Because it will have to be done more than two and a half times and done all the time completely
And since the house itself will be expensive - all equipment will have to be taken expensive
Smart apartment, by definition, cannot be
Because, while there is no ISO - I'm not sure that the industry exists Even despite the
wild
hype

Answer 10

Respect, a fairly competent article - otherwise you read one "water".
1. Machine learning - do not tell my slippers. If a house costs 10 lyam, then why do I need AI for a hundred?
2. The whole problem of "smart domists" is that they are not good at interconnection patterns - an individual approach is practiced everywhere.
3. A single organism-complex has a significant flaw - it is a person, that is, the end user.
4. And yes, you are not the first and not the last to talk about unification - it will not happen until there is an industry leader.