Comment by brk | Hacker News Reader

Comment by brk | Hacker News Reader

This is interesting. I've been involved in home automation for about 15 years now, and HVAC controls are one of the top things to control in most systens.

Occupancy sensing is also extremely difficult to do passively in any kind of reliable scenario, especially when your occupancy sensor is limited to a single location in the house.

I've frequently seen two broad categories around regulating HVAC systems: 1) People keep a fairly regular schedule, and a standard 7-day programmable thermostat works sufficiently.

2) People keep a very non-regular schedule, and a standard programmable thermostat doesn't work, nor is there any inferable pattern to their home/away schedule.

In case #2, it is usually more beneficial to get home/away status from something with a more direct output eg: a burglar alarm system, vehicle presence detectors in the garage (photobeams), status of lights, etc.

I'll be very curious to see what the real-world reactions are to the Nest thermostat. If they can come up with some truly creative solutions it could be a really cool device.

Why not setup a system that detects the presence of your Bluetooth phone? Perhaps even look for your phone on the wireless network if a sensitive detector is not sufficient?

It would not be a stretch at that point to have a simple website served from the device that let you see the history and choose some basic night/day setback points.

It takes some time to raise/lower the temperature of the house depending on the outside temperature so learning the rate at which the house warms/cools is required to reach the desired set point at the requested time. Including the local weather forecast as a sensor and being able to see a graph of temps and system demand cycles would be really nice.

I would be willing to carry something-er-other in my wallet which would alert my home automation system (which I don't have) to my presence/location in my home. I usually leave my wallet in our bedroom when sleeping, so it would still be accurate at night. My wife might be willing to wear a small ring or something like that, but it would be a harder sell.

The benefits to the phone, you already have it and keeping it charged and in general not forgetting it at home are more or less taken care of.

If I were considering options for additional units to purchase for determining occupancy (perhaps the kids do not have cellphones yet) then a small zigbee type mesh sensor would be ideal. This type of solution would give many options for those with zone based heating and cooling and would lend itself to even more pretty visualizations.

Edit: Getting a post right the first time from a phone is harder than it should be.

I had something a while back... A little script thingy I found online that would lock/unlock my laptop (MBP) based on if it was paired via BT to my phone. The problem with BT occupancy sensing is that the range is kind of limited, so you need several sensors, and the pairing dance that goes with that, it's kind of a hassle.

Why not simply have it get the data from an online API connected to something like Google Latitude or Apple's Find My Friend system? Then you could even have it make decisions when the users are much farther away, e.g. "he's 100 miles away, just keep the house above freezing" or "he's come back within 10 miles, return the house to comfortable levels."

I know some folks who have a heater in their winter cabin they can activate remotely (via text, I think) to turn on before they arrive for the weekend. An automated version of that would be even better.

Bluetooth drains the battery too quicky. An app running in the background with gps pings would be better.

From what's on the website it looks like they're trying to learn largely from user feedback of the "I wanted it cooler/hotter" variety. Agree that this will be a problem in case #2, without good occupancy sensing, because the prediction problem is inherently impossible to do well with only time-of-day-and-week if there isn't much regularity.

I would guess the real target audience is people who fall into #1 but never get around to programming their thermostat, and maybe don't even have an accurate self-assessment of their home/away or sleep/wake patterns. In particularly warm or cold climates this might also include people who make wrong guesses about latency, e.g. how early before they return the heater needs to kick on for it to be comfortable by the time they get home. I could see it being useful in that case: there's a regular enough pattern to infer a reliable heating/cooling schedule, but the person doesn't want to, or isn't able to manually program that pattern.

On the website "Activity sensors Nest's activity sensors have a 150° wide-angle view so Nest knows when to set itself to Auto-Away."

At $249 though it needs to make some sort of sell that I can recoup more than those costs.

That's why I like my OmniStat with my OmniPro security/automation system, it can change temp based on security system state (away, night, etc.) And if you still wanted occupancy based learning, it can be done with your security interior motions and Omni programming.

Also the Haiku iPhone app for my Omni just added changing flags in the OmniPro based on iPhone location triggers, this would allow proximity tstat changes if you really wanted that.

http://homeauto.com/Products/Omnistat/Omnistat2Products.asp

I have an OmniProII system, but have scaled back what I am using it for. In many cases though it makes a very nice security, HVAC, automation and audio platform.

#2 today could be solved with smartphones that relay their location?

ZigBee wireless sensors can be built rather cheaply if the intelligent components are centralized on the primary controller.

It would be great if this also supported Z-Wave, Insteon or ZigBee. Z-Wave would probably be more advisable since ADT and a few others are jumping on the Z-Wave protocol.

It supports Zigbee ("Show detailed Specs" link at the bottom of http://www.nest.com/inside-and-out/index.html)

They have an iPhone app. If that can pull GPS coordinates then they can know where at least one person is.

Reply on news.ycombinator.com