I've played around with Home Automation for more than 2 decades now, and have many automation systems in my home including lights, heating, curtains, blinds, CCTV, and integrated security alarm and have Sonos around my home for music. I have over 20 motion sensors around my home connected to my home automation systems and I've experimented with many technologies, systems, and applications in my time, and have a good idea of what is practical and what isn't.
One of the common things I come across is people saying "wouldn't automated control of your home based on motion be cool". Often some of the ideas are wacky or just half-baked. Mostly they are from people who think it sounds "cool" and haven't thought it through past the initial concept.
Recently I was pointed towards this Kickstarter project which takes a half-baked idea and tries to sell it to the sort of gullible people who are more interested in "cool" than practical or useful.
Firstly, I should state there is nothing new or innovative about what they are doing. Systems for home automation based on motion sensors have been around for two decades. I've had one in my home for over 17 years. I can use motion to control my lights, curtains, blinds, heating, and even the vents in my conservatory. I could also use it to control my music.
I haven't done most of these things because the concept is fundamentally broken in most of these cases.
I should point out I have tried most of these things, but I found they really didn't work. I stall have motion based automation, but it is restricted to heating and, to a limited extent, to lighting.
I will reiterate that this is not new. You can get motion based home automation based control systems from many, many vendors. Here are some: Cytech, LightwaveRF, Genius Hub, Owl, X10, Fibaro, Vera, Honeywell, Tado, and many others. To get a flavour, look at a typical Home Automation website like Vesternet. Many of these systems are inexpensive and extensible (and a lot more practical than a platform based solely on motion sensing, like Trajectio).
But more importantly, what they are proposing is largely impractical, or of limited practical value. In my experience motion-controlled home automation only works in a limited subset of applications. Here's my explanation of what works and what doesn't, and why:
Heating/HVAC
Heating/HVAC is a great application of motion based control in many ways as it's a high-hysteresis, low-information, high-predictability system.
By "high-hysteresis" I mean by that is that it takes a long time for a room to heat up and cool down so you generally don't want it turning on and off every few minutes. It's more sensible for heating to turn on for longer periods of time. It's also sensible to delay turning on heat until a number of minutes of motion has been detected: you don't want to turn the heating on for an hour in a room because someone popped in to grab their keys off the table. Delaying the heating start isn't also going to make much difference to people's comfort.
By "low-information" I mean that the heating system really has two states for a given room: on or off. Of course it can have a temperature, but that is usually set in advance for the room. In fact my heating system automatically adjusts based on preset separate temperatures for "occupied", "unoccupied" and "night".
By "high-predictability" I mean it's relatively easy to assess the required outcome based on the input: if people are occupying a room for a given period of time, they will want it heated (or cooled).
Of course, it depends on the room and people's movements, and I have found, in practice, you have to "fine tune" these timings to get close to the ideal. My heating system also learns over time, so if it sees motion in a room at the same time every day, it will start to pre-empt that and turn the heating on before time.
Also the outcome is not specific to the person who occupies the room: in general the heating is required to be turned on regardless of who is in the room.
Lighting
Lighting is less easy. It is a zero-hysteresis, low-information, low-predictability, moderate personalisation system.
"Zero-hysteresis" because lights go on and off instantly
"Low-information" because, in general, lights are on or off, but in some case you can also have dim levels
"Low-predictability" because it's not easy to know what is the best action to take based on the input. I quite often walk into neighbouring rooms and don't turn the light on because I'm just grabbing something from the side. It wouldn't be terrible if the system turned the light on for me, but it would be a waste of energy if it then kept the lights on for (say) another 10 minutes every time I went into a room for 30 seconds. In fact, I originally tried automating the lighting for most of my downstairs rooms and found that a lot of my lights would actually be on most of the time due to transient traffic from people and pets. I also found that, even when I spent a significant time in a room, the lights would turn off at random intervals because I wasn't moving enough to keep the motion sensor happy. And, unlike heating, you can't sensibly delay it turning on.
Trying to strike a balance between keeping lights on too long, and not enough is a far hard problem than the heating one, largely due to the difference in hysteresis and user expectations between the two systems
And different people have different expectations and requirements. If I come down for a drink at 4am, I don't want to be dazzled/surprised by lights coming on automatically. My daughter would probably injure herself if she didn't turn on at least one or two lights. When my wife goes into the room, she likes every possible light turned on. I often prefer just one.
Music
Music is close to impossible to do except in very simplistic cases (small houses with only one or two people with inflexible routines). It is a zero hysteresis, high-information, almost zero-predictability system.
"Zero hysteresis" because, with a system like Sonos, playback is pretty much instant.
"High information" because there are almost infinite choices of what the system could play when you factor in radio stations, playlists, and music services.
"Almost zero-predictability" because it's pretty much impossible to know whether the input even means that music should be played, yet alone what source to play, which playlist, etc. even in a home with only one person. With multiple occupants, it is literally impossible to predict whether, what, where, how loud, and how long to play music with any degree of accuracy.
Even if you take the view that you only "follow" occupancy around when music has been chosen and started, this is still highly unpredictable: if I'm at home alone, I might be quite happy to start music in the kitchen and have it follow me into the living room. When my wife is at home watching TV, I don't want that at all. Especially as the system cannot distinguish between occupants, so my wife shifting position on the sofa would, in many cases, be treated the same as me walking into that room.
Frankly the number of cases where music following occupancy would break and remedial action would be needed are enough that it's less effort to just manually group zones where needed. When I looked at whether I could, meaningfully, trigger Sonos using motion I found pretty much every predictable scenario I had was better implemented using alarms.
Summary
Save your money, don't invest in this nonsense. It's not new, it's not innovative, and some of the concepts are fundamentally impractical.
No comments:
Post a Comment