Editor’s Note – This is the second piece by fellow journalist and Geek Marty Winston as he chronicles the creation of his home of the future with today’s best technology.
This week, we go deeper geek.
It starts with some disappointing technology many people experience that’s supposed to turn the lights on when somebody is in a room and off when it’s empty. The least expensive (which may explain why it’s the most popular) technology for doing this is passive infrared (aka PIR). It is, alas, a motion detection technology, so in its eyes, if you’re not moving, you’re not there. If you ever sat alone reading or working at your keyboard in a place like a conference room and had the lights turn off as you sat there, you know the flaw in PIR.
I had to come up with a different approach for my autonomous house project (http://40yearhouse.com) and made a propitious decision to identify warm bodies as a more foolproof approach to identifying occupancy versus vacancy. That led to quite a harvest.
CAP: CEILING AWARENESS POD
One of our underlying principles is to avoid the “wall measles” of little white plastic blobs stuck all over the place; these are the metaphorical acne of current home automation products; we believe any automation presence should be no more visible than a smoke detector. Another of our principles is to ban batteries – all batteries eventually fail and people are always terrible at babysitting batteries – and when a problem arises, it’s extremely unlikely that a homeowner could ever trace it back to a single device and a dead battery within it.
We recognized that detecting warm bodies in a room would work best with sensors placed overhead, so we started our thinking in terms of a standard 2-gang electrical box in the ceiling and things we could put up there. That ended up (we’ll tell more of its story in future installments) as more than 5 dozen sensor readings, a small amplified speaker, a connection to a magnetic switch at the door to the room, a relay to control an adjacent ceiling fan and a small computer, the size of a deck of cards.
GRIDEYE AND RASPBERRY PI
Our house plans call for 30 of these CAP devices, each able to cover up to 16 feet square. The most amazing sensor in each is a Panasonic GridEYE, which can digitally report analog temperature readings in an 8×8 grid of 2-foot-square thixels (thermal pixels). In addition to these 64 thixel readings, it also provides an on-chip thermistor that reports what it regards as the background temperature. It can report those readings 10 times per second.
We originally planned on using one of the low-energy wireless protocols to relay those readings but, to make a long story short, it was too much data for them to reliably communicate. We needed something to send that data over a wire. (Why not WiFi? Because we have too many endpoints for a WiFi router to sustain as simultaneous, always-on connections). Our wired solution turned out to be 100 Mbps Ethernet, and the cheapest way to grab the data and send it over an Ethernet connection turned out to be the $36 Raspberry Pi Model 3 B.
And that was a springboard. The Pi is a surprisingly capable single-board computer. Using it meant we could do a lot of in-place analysis and deal with more features in one place.
And its first task is to build a heat map.
PLUCKING WHAT’S HOT FROM WHAT’S NOT
We arbitrarily set 90 degrees as a temperature watershed: anything below 90 is considered to be a room background temperature and anything above is interpreted as a warm body (or worse). We know that healthy adult body temperatures can start at 96 and, in clothing, may read a bit cooler; the light-bulb-bright heat source for humans is our heads, but we have to allow for wigs. We know that children tend to be warmer than adults, infants warmer than children and pets warmer than infants, so we created some scoring ranges the readings might represent. Above those temperatures and below 250 degrees flags an obvious trouble zone; above 250 gets recognized as a pre-combustion fire danger. (Actual combustion tends to
involve temperatures above 400 degrees, but we rely on 14 interconnected Kidde FireX smoke and CO detectors as our primary signal of actual fires).
Any one CAP can essentially cover something the size of a guest bedroom. Each CAP gets individually configured with lists of thixels (little squares underneath) that we should treat separately. So, for example, if a room is smaller, we can list the little squares that don’t represent floor space and ignore them. If a room has known hot spots, like cooking surfaces or a fireplace, they get listed as permissibly hot spots. Areas near doors are important for other reasons, and we list those. So while every CAP can report on 64 of those squares, some will pay attention to fewer.
Like a lot of programmed computer activities, the Pi steps through all of those thixels. Any warm body thixel goes into a warm body map (actually a data set of location and reading). All the other thixels get treated as background temperature.
BACKGROUND CHECKS
Just because it’s the background doesn’t mean it’s unimportant. During each of those 10 scans per second, the program takes special note of the highest and lowest background temperatures. It those are separated by more than 10 degrees, it’s a sign of trouble. The system initially responds by turning on the ceiling fan, if there’s one there, to stir the air; that’s why there’s a relay in the CAP. If, after a reasonable time, the wide variation continues, the system sends the occupant an e-mail alerting to a likely problem with HVAC systems or insulation.
That won’t often be the case. Usually, it will just average the background temperatures and report that. We’ll get into what happens as a result of background temperature reporting in a future column (it’s very cool).
BACK TO THE FOREGROUND
Those non-background temperatures are our warm-body temperatures; are there any within any one CAP’s view? If not, we know a room is vacant; if so, we know a room is occupied.
Sure-footed occupancy/vacancy information is a great asset when it comes to saving energy. It lets us know when to turn lights on or off. (One reason to pay special attention near doors is so we can anticipate transitions from one place to another). There’s more – here’s a more complete CAP recap:
Lighting controls: All of the lighting in this house is done with LED fixtures, and all of those get controlled by products from Synapse Wireless. We worked with them to find ways for our automation to talk to their automation. Knowing where people are and where they’re headed is part of the decision-making in the lighting controls.
Comfort controls: We essentially create two thousand remote readings for a thermostat, which alone might be a little help. It helps more that we sometimes quite deliberately lie to it. If the house is set at 70 and there’s a guest bedroom that’s not being used (but only with its door closed, or this would be a waste), we can treat any temperature from, say, 60 to 80 
degrees as if it’s 70, meaning more comfort where people are but less energy overall. It’s even better because we get air to each room tuned and trimmed by motorized vent shutters and a very intelligent hub from Ecovent Systems.
In case of fire: This is one of the coolest features of the house. The “vanity” (house number) panel at the front door is really a digital sign. If the house detects a fire, one of the things it does is to put a display on that sign of a floor plan of the house showing the position of every adult, child, infant, pet and hot spot. When firefighters arrive, it directs them to where they can be their most effective most quickly.
The house does many other things if there’s a fire. It phones in a voice call to the fire department, turns off the gas line right after the tap to the generator, turns off every fan in the house, gets the Ecovent shutters to all close, unlocks all the exterior door deadbolts and places the cameras monitoring those doors on full-time record.
ADDENDA
The CAP has a Sharp proximity sensor so if there’s a ceiling fan that isn’t powered on and its blade is blocking the GridEYE view below, a quarter second burst of power to the fan (through the relay) can nudge that blade out of the way.
An amplifier and speaker let each CAP act like a local doorbell chime, and then some. We have a library of MP3 sounds stored there to separately and distinctively signal when mail arrives, when vehicles enter or leave the driveway, when a car enters or leaves each of the 3 garage spaces, when there’s a truck outside the front door and so on. Using a Raspberry Pi to play the sound means we can be selective – no need to play to an empty room, or to a kid’s room or guest room when they’re sleeping, or inside any bedroom at night.
There’s also a connection to a magnetic reed switch in the room’s door frame to confirm when its door is closed, an ambient light sensor and a relative humidity sensor.
But for all of this, a CAP is just an intelligent endpoint; the nervous system of our house has many others. A visit to http://40yearhouse.com can give you a sense of others – and so will we, right here, starting next time.
The post Smarty House Part 2 – Tracking Warm Bodies appeared first on Tech50+.