Comment by kinsomo | Hacker News Reader

Comment by kinsomo | Hacker News Reader

Parent Full thread

> I have this idea of a system I would like to have in my house. It contains cameras in every room that are constantly watching where people are and relaying the coordinates to a central server. That server makes decisions on if lights should be on or if A/C should be running in that room. But I would never buy this system. I would have to make it myself. I am hopeful that open source software and hardware can produce individual components that I can trust to piece together.

You don't need cameras for that, just motion sensors.

> That server makes decisions on if lights should be on...in that room.

You don't need a server for that, just a motion-sensing switch. They can be totally offline. My office has them to shut off lights automatically when a conference room becomes unoccupied.

https://www.homedepot.com/b/Electrical-Wiring-Devices-Light-...

It can get annoying if you're not moving much, but the key point is: no camera or server required.

> That server makes decisions on...if A/C should be running in that room.

Ecobee thermostats can already do that using motion/temperature sensors. They run fine without an internet connection.

https://www.ecobee.com/room-sensors/

But then how can I sell the analytics data for billions of dollars? /s

Might be cool to add thermal sensors to that, so the lights don't shut off because you're not moving enough.

Add in voice, image recording to get gov income as well.

That would still mean that the lights would go off if my boss was the only person in the room.

That's alright, he doesn't really need the light anyway.

That's the main problem with motion activated office lights tbf, especially when you're doing software development for example - not enough motion to keep them on.

Eh, it encourages me to get up and have a stretch periodically.

That doesn't work in a toilet stall.

Every time! The coffee shop by my old house had motion sensing lights in the washroom, and every damned time I would be plunged into darkness half-way through my visit... Urgh.

May I suggest some fiber?

Our company has this in the toilets, when it goes dark you just wave your hands

Ours does too, but there's no sensors inside the stall... You can wave all you want, there's no way to turn the lights back on without opening the stall door.

Thank goodness for phone flashlights!

My problem with the motion activated office lights (at least the ones at the wework my company is located) is that you can't turn them off. The button appears to turn the lights off and disable the motion sensor for some (short) period of time, then movement turns the lights back on again. Very annoying.

My office has a piece of paper taped over the motion sensor on the roof for that reason.

could be fixed with sensors in all the seats, or on the floors where people with standing desks are

I wish more offices would install motion sensors facing keyboards from above, tune-focused for finger motion detection.

> I wish more offices would install motion sensors facing keyboards from above, tune-focused for finger motion detection.

A new dystopian developer-productivity metric.

Erm, could you not also just install a keylogger?

The NSA could just give you access to theirs.

I'd also like them to dim if I'm holding a glass of wine and/or move in a particular manner.

Edit: typo

Haha, and play some jazz tune maybe?

If you're staying that still, you likely don't need room lights on. Why not just use an area/task light if you're performing a stationary task?

For people with glasses the reflections can be pretty bad. In addition I require bright cold-white neon light or I fall asleep.

By the way, German workplace safety regulations require a certain minimum of light at a workplace.

OSHA also requires a minimum light level in the US but I believe it only pertains to hazardous working environments (like on a production floor in a manufacturing facility for example). Either that or the enforcement in white collar environments is so weak that literally no one cares to follow the law which is surprisingly common with reagrd to many workplace regulations.

commodity motion sensors are thermal sensors[0]. Perhaps a FLIR[1]?

[0] https://en.wikipedia.org/wiki/Passive_infrared_sensor [1] https://en.wikipedia.org/wiki/Forward_looking_infrared

FLIR for turning off a light in a room seems like a ridiculous amount of overkill. I don't know numbers off the top of my head but I feel like at that point you're using more energy for FLIR than the light itself. Not to mention the cost of a single FLIR sensor is probably pretty high.

I think clever placement of traditional IR sensors is the answer here.

These cheap 8x8 sensors may be useful[0].

[0] https://www.adafruit.com/product/3538

Could be a problem if you plan on sleeping, but you could some rules to get around that

One reason you might want a server is to have more expressive power than "if someone literally enters the room, turn on the lights". You might want to turn on the lights and run the AC a little before you predict people will arrive, for example.

> One reason you might want a server is to have more expressive power than "if someone literally enters the room, turn on the lights". You might want to turn on the lights and run the AC a little before you predict people will arrive, for example.

Maybe, but that seems like a lot of work for little gain.

You probably don't actually have that much control over the AC in your rooms unless you have zoned heating and a lot of zones, regardless of how smart your sensors are.

Intent is a very important factor when answering the question "should the lights be on," I don't think you'll be able to predict that. For instance: if movement is detected in the bedroom at 2AM, should the lights come on? The answer is: a very strong maybe.

> Intent is a very important factor when answering the question "should the lights be one," I don't think you'll be able to predict that. For instance: if movement is detected in the bedroom at 2AM, should the lights come on? The answer is: a very strong maybe.

You would be collecting data from many sources in order to predict intent. This is why you need a centralized server.

> You would be collecting data from many sources in order to predict intent. This is why you need a centralized server.

Please explain, exactly, what other sources you would collect data from and how the central server would process it to determine if I want the lights on in the middle of the night.

IMHO, determining intent in this scenario is impossible without...

1. a mind-reading sensor, or...

2. an explicit user signal, such as a button-press or command.

The only realistic option is a user signal, and most of those options obviate a lot of these prediction ideas.

I think there's a lot less practical value to having a "central server" controlling everything than you seem to assume.

> Please explain how, exactly, what other sources you would collect data from and how the central server would process it to determine if I want the lights on.

The more data, the more smarter. In fact, the only difference between a thermostat and the human mind is the number of datas. This is because the Law of Averages predicts that half of all datas will be relevant.

> The more data, the more smarter.

There are several important caveats to this statement.

> In fact, the only difference between a thermostat and the human mind is the number of datas.

This is an oversimplification that borders on dogma.

> This is because the Law of Averages predicts that half of all datas will be relevant.

An interesting interpretation and application.

I really really hope this was a joke, because it was incredibly funny if so and terrifying if not.

> Please explain how, exactly, what other sources you would collect data from and how the central server would process it to determine if I want the lights on.

Machine learning navel-gazing is the new "throw a start-up at it", so I'm guessing the answer to this is going to be "if we have enough data..."

> Maybe, but that seems like a lot of work for little gain.

That's how I feel about Alexa.

While I haven't had the urge to get any 'smart speaker', I have a hunch that the current hype* around these devices comes mainly from the novelty of being impressed with voice recognition.

As someone who doesn't have a voice for radio, I much prefer being able to interact through dexterity. Besides, I'm very picky when it comes to which specific track of music is to be played (out of various similarly named pieces of music).

* I don't think take-up is quite as big as the tech world currently makes it out to be. People in my circles don't really have that kind of disposable income and are prioritising other purchases first.

So yeah, Alexa is a lot of extra literal work for little gain.

If you don't have multiple zones you can still effectively have zones by opening/close the registers/vents. This will prevent hot or cold air from entering the room. However, it will also increase duct pressure and that could be somewhat problematic. Either way, "smart vents" are on the market. They're ridiculously overpriced though and one could hack together an alternative for under $50 per vent (source: personal experience). Probably less if they are clever. Of course then there is the problem of running power to the vents or having sufficient battery. But that's an exercise left for the reader.

Your solution is simple. However, it doesn't allow for as much modification as the original poster's idea. With his idea it could be modified in so many ways to add functionality because it wouldn't be limited by the technology. The solution you supplied will eventually be limited by the technology if the original poster wants to add other functions.

Sure, but what is the value gain/opportunity cost between a simple, two hour installation that achieves a majority of the desired outcome with nearly rock-solid stability vs. sinking a massive amount of time into a bespoke and likely fragile system?

> However, it doesn't allow for as much modification as the original poster's idea.

My main point was the original poster's idea was probably focusing on the wrong kind of sensor for what he wanted to do. He could still network a bunch of motion sensors.

Also, I'd dispute the idea that my proposals were less "modifiable." If only because they're far easier to implement and a couple of orders of magnitude cheaper, so it's practical to replace if more capabilities are needed.

There may be a market for something between a standard one-bit low-rez motion sensor and a full color TV camera. Maybe a 16x16 pixel IR sensor with a fisheye lens and a puny CPU that reports an approximate number of people in the area, for HVAC and lighting control, and security.

Reporting an approximate number of people is not a trivial task, even with a real camera. Depending on what exactly you mean with "approximate", of course.

0, 1, a few, many. Just enough to tell you how much to crank up the HVAC.

> 0, 1, a few, many. Just enough to tell you how much to crank up the HVAC.

I don't think the number of people in a room will give you meaningful information to tell you "how much to crank up the HVAC." Also, the HVAC systems in most homes aren't capable of even cranking up the HVAC in a particular room.

True. The market for this is schools, offices, and hotels, where the people load changes drastically and the HVAC has to react to that.

This is what the "Internet of Things" should be doing, but seldom does.

YAGNI. You can get to a 90% solution today at 10% of the cost and effort. If, later, you want to extend the system you’ll not only have learned a lot about the operations and failure modes of the current system, but hardware purchased later will be cheaper and may support even more functionality.

http://www.danielsen.com/jokes/objecttoaster.txt

> The second advisor, a software developer, immediately recognized the danger of such short-sighted thinking. [...] "A toaster that only makes toast will soon be obsolete. If we don't look to the future, we will have to completely redesign the toaster in just a few years."

If you buy occupancy switches instead of vacancy switches, the light will stay on until you actually leave!

Reply on news.ycombinator.com