My latest cool DIY automation idea

It's about as trivial as a three body gravitional problem.

The basic problem is most sensors will not see a person sitting quietly in a chair as motion and they'll turn things off. It's called the automation "wave" because people then sit there flapping their arms to get whatever the MD shut off to come back on.

Figuring out who's where and doing what in a house, short of making everyone wear transponders, is very tough business. For starters, you need rules like:

If HALL MD NOT TRIPPED AND LR MD NOT TRIPPED (it's likely no one left the LR even though there's been no motion detected for 5 minutes otherwise they would have tripped the hallway MD on their way out!) THEN DON'T TURN OUT THE DAMN LIGHTS! :-)

A system at least has to be smart enough to "realize" if you didn't trip any sensors outside the room, you probably haven't left it no matter what the timer and PIR "thinks."

I can't think of anything with a lower spousal approval factor than a Robohome brain turning off the lights or sound because someone was sitting quietly minding their own business.

The perfect solution would probably encompass microwave detectors, RFID sensors, IR beams, PIRs, pressure mats, 1000+ lines of code and a fast CPU to figure out where people *really* are in a house. Anything less is going to get it wrong a lot of the time. So much so that it becomes a serious annoyance as people get used to flapping their wings like insane geese.

Ironically, I see some great parallels in game software design and occupancy sensing. You'll need variables for each "player" or occupant indicating current positions and a set of rules governing how they can move from room to room. Sounds like fun. I'd like my HA system to have a "God" mode.

-- Bobby G.

Or simply mount each room's floor on four load cells, greatly reducing the coding complexity.

I leave the constructional details as an exercise for the reader.

Then you have it say things like, "Good evening, Gladys. At least I think this is Gladys based on weight estimation, though you seem to have put on a few pounds recently."

------------------------------------- Dean Roddey Chairman/CTO, Charmed Quark Systems

That would be a "strain gauge" in more ways than one!

-- Bobby G.

Yup. But if everybody _were_ RFID'd, it seems to me that would suffice.

You mean locking the kitchen door after 10PM if you weighed more than TargetWeight + n kg unless you logged in f(n) x m minutes on the Nordic Trac ? ;-)

More seriously, why is RFID by itself intrinsically inadequate for simple (binary) localization (Is given person in room - yes/no)?

( A simple turnkey RFID solution exists now for at least one popular home automation software, namely Homeseer. Homeseer addons can also do rudimentary localization using a generic Bluetooth cell phone.)I have no experience with either. )

Sophisticated systems with multiple sensors can locate an RFID tag to within centimeters in 3-D space in real time. Do we need better than a 4-D definition with centimeter and second resolution for occupancy _detection_?

This time-space resolution should also be adequate for calculation of velocity vectors (how fast and what direction is an object/person travel ling) that in turn can be used in a predictive mode, namely where is the body headed next.

FWIW, RFID is advanced as superior to GPS in some situations for first-responder emergency needs by these folks from the National Institutes of Health:

This url also has many useful references and links to RFID indoor use, localization techniques, pedestrian navigation, applications of Zigbee, phone-based localization among other topics.

... Marc Marc_F_Hult

All good points, one data point I didn't provide: We have a 4yr old and a

2yr old child. Literally the only time there's no movement in a room we're in is when we're watching TV after they go to bed. My wife jokes that the soothing blue Bosch BlueLine motion light is the real Big Brother.

In addition, i'm not looking to turn it off if we won't be in the room for only a short period of time.

Here's the additions to the plan:

- If a DVD or the TV isn't on, then do those rules.

- Expand timeout to 15mins

- Create a "turn God Mode on/off" button that can be accessed via the inwall touchscreens or a PC, perhaps even IR that also disables the rules. That way, if there's an activity we're engaging in that is prone to false positives, the whole thing can be suspended.

Perhaps I'm just naive, but I really do think that would expand accuracy to >90%, at least in our house. The "God mode on/off" button would tackle the other 10%, and disable it.

Then again, I may report back here that it was a colossal failure. Time will tell. Hey, at least this is an option available to me, whereas pre-HA it was not.

That certainly differs from my environment!

That may make it all the more jarring when you do have quiet time and the damn lights go out. I'm still looking for ways to solve the problems that real "occupancy sensing" entails. I hope I haven't sounded critical of your effort. It's just that you reminded me of me when I first attempted to deal with automated lighting. My experiment was doomed from the start because I was using way too many X-10 PIRs. It resulted in powerline collisions that turned off entirely unrelated lamps. That turned out to be the end of automatic room lighting for us (moreso for the wife than for me), at least the kind that was controlled by X-10 PIRs.

My wife agrees with you. She finds that the automated turning on of lights are most helpful when she's got her hands full and is walking into the laundry room or the bathroom. However, if it were up to her, they would never turn themselves off automatically. She just sees automatic turnoff as catering to my bad behavior in not turning off lights when I leave a room.

When I wrote about "God mode" I was kidding but now I realize that it's actually a good idea the way you've described it. I think my wife will like it even more.

You're not being naive. It's just that battle plans rarely survive contact with the real enemy. Anyway, it's all pretty subjective. I'll bet with a little fine-tuning you'll get things pretty much to where you want them. I'll probably do IR beams across each doorway in the next home as one of the more reliable ways of detecting movement between rooms. I'd use RFID tags but until I find one made by Cartier, I don't think I'll be able to convince my wife to wear one 24/7.

I'm working on ideas for multiple sensor "nets" so that there's always some way to verify the information one sensor sends with that of another. I bought a big roll of pressure mat material that I'll be installing on each side of a door in the new house to hopefully give me information on an occupant's direction of travel as well as their presence in a room. That information, along with PIR and IR beam data, should be able to conclusively determine whether a room is occupied or not. There should also be enough redundancy so that the system can operate reliably if one or more sensors fails or has failed to "see" the occupant.

Good luck and be sure to let us know how it all works out!

-- Bobby G.

A Fujtisu 3400 tablet PC w/touchscreen. The first one cost me $250 6 months ago, just picked up a 2nd one for $160 [both off eBay]. It's aesthetically "ok", responsiveness is decent but I needed to create bigger buttons. It's passive touch, so you can use your finger, but my wife opted to put a ballpoint pen with the tip removed on top of the frame as her hands are always messy since it's in the kitchen.

I got pics of the whole install project here, haven't updated it since I buried the 12VDC line. Pretend the wire isn't there in the last pic, and you've basically got it.

This is the interface set that I use with it. You can see how the buttons are a little bigger than what most would use.

Now that is cool.

"E. Lee Dickinson" wrote in

I wonder if there's enough space to mount two LEDs and two detectors? I was going to use the pressure mats to make direction of motion estimates, but a dual IR beam would work as well as long as I could tell which fired first.

Chronoton detector.

-- Bobby G.

"Schrödinger's cat-litter box. If you don't observe it, you won't have to change it."

What are you using for touchscreens?

Bobby,

I stumbled across a solution to this at the airport, just the other day. Pulling in to the parking garage, there was a sign that indicated the number of available spaces on each floor. As you all know, I'm a geek with a touchscreen in my kitchen cabinet, so naturally I had to figure out how this worked.

I stomped around on the concrete for a while, trying to find the pressure sensors in the slab, looked above for the laser range finders pointing down above each space, and checked for underpaid workers with tally sheets.

Finally figured it out, and it was so elegant. Every entry and exit to the parking garage has two IR beams, spaced six inches from eachother. Whichever beam is broken first indicated the directon of travel. Car goin in, add one. Car going out, subtract one.

Same would work in a door jamb, put it about 3 feet up so it misses pets, but detects (many) children, and doesn't count adult legs as two people.

Of course, this can't account for people jumping out of windows, beaming up, or falling through transdimensional rifts. But after all, this is DIY... :)

Simple concept and I manage to make a narrative out of it. Sorry. :-P

I dig that. I was just about to drop some money on a few Harmony remotes. I might have to get some tablets instead. :)

I'm still pretty proud of my kitchen cabinet screen

but need to move the computer out of the kitchen cabinet. I'm having heat buildup problems and space issues.

I found a great balun online that carries VGA, USB, and stereo audio over one line of Cat5.. for the low low price of $700!

That is one seriously nice woodworking job. I recently cut a hole in the floor of my media closet, which is directly above my crawlspace, so I could vent it. I put in a heat register grate for the finish.

Let's just say that I was very thankful the grate had 1" of overhang on either side.

For the full details:

Yeah, that's why I LOVE those 3400's - for $200 shipped, plus a $8 shipped hardwired or wifi card, I get that wallmount. Sure it could be cleaner, but it's $200. The wife uses it constantly, and actually shows it off to her friends when they come. Now that's adoption.

When I hit the lotto, i'll get those TouchTronix screens/UTMA jobbers, until then i'm quite ok with this.

There's another solution that you might want to consider. Visonic makes a "SPY 1" motion detector that can be mounted in a wall or the ceiling close to the door. These are PIR detectors in a very small package that fits into a hole in the sheetrock.

The advantages of the SPY are (1) it's much less expensive than most pulsed IR beam detectors; (2) it's physically smaller than all but one of the active IR beams used for security and (3) the SPY only entails a single wired device per location, whereas most active IR devices have both an emitter and a receiver.

"E. Lee Dickinson" wrote in

That looks suspiciously like a mini-ITX next to that colander. :-)

I asked my wife where we should mount an LCD display in the kitchen. Her first answer was that the microwave should be able to switch between clear window and LCD screen. While that would be nice, I don't think I'm going to be able to do it.

The second choice was behind the sink where the backsplash is so that she can watch TV while working at the sink. I want a screen I can see from the breakfast table but I also would like one that I can see from near the fridge so I can enter items that need reordering.

I am definitely mounting the PC outside of the kitchen area because all space there is spoken for and even a Mini-ITX format takes too much space.

-- Bobby G.

I would mount one of these

opposite the door and range find the object as it moves through the doorway.

Getting closer = entering, getting further away = exiting.

Just an idea.

Cheers

Rohan

A pretty neat solution and certainly something to consider in addition to all the other technologies.

The problems of occupancy detection are in some ways similar to aircraft radar detection (sky occupancy, if you will). Single targets are easy to identify. It's when two targets merge and then separate (flight KAL 007 for example) that trouble begins. I don't have any figures to back me up, but I feel that a two IR beam detector at each doorway would be less confused by a room full of people than a rangefinding device.

There's no doubt, however, that even two IR beams could be confused by two people walking through the doorway together. That's where pressure mats on either side of the door would provide the definitive answer. Even if the IR sensed two people in the doorway, the mats would tell the tale of which way they went. No weight detection on one side meant they both went the other way. I do see rangefinder technology as a further way to test the assumptions of occupancy. The units you pointed to are nice, small, have very low power consumption and are fairly cheap (25 to 29 bucks) and I would probably use them where my ultra-cheap sensors were problematic. The look especially useful for robotic designs.

Gotta love the builder's motivation: "my daughters robot was stuck on a wall."

Any idea what happens if the units are placed where they can "hear" each other?

(Answered, actually, by the FAQ:

at their very simple but very well done website)

The designer details several ways of using multiple sensors without them interfering with each other. He also points out the problems facing ultrasonic detectors and "soft acoustical targets" like human beings.

That discussion leads me to believe that double beam IR doorway sensors are still the most reliable ways to detect exit and entry and therefore occupancy. The only hard part is designing a circuit that keeps the two LEDs from interfering with each other. That will probably require a design similar to theirs that reads each sensor "round robin" style.

I thought of doing it cheaply by have a low sensor and a high one, but I really wanted both sensors at the same height so that I could reliably sense motion. At the same height, one would always be "eclipsed" for a very short time before the other. A simple latching relay circuit could detect direction based on which LED fired first. With detectors at different heights, you could mess up the direction if someone's leg crossed at the exact time that their torso did. At exactly the same height, that wouldn't happen.

The only unanswered question remaining about the Maxbotics ultrasonic units: Will they attract bats? ;-)

-- Bobby G.