I want to rig up a system in my house that triangulates my position

>>1092285
what the shit exactly are you trying to do

>>1092287
Trying to integrate video surveillance, smart electrical systems, and pretty much everything else with a plug into one system that activates/deactivates systems based on what room I'm in.

I suppose I should elaborate. I have easy access to a leica p20 laser scanner, so I'm thinking of creating a perfect virtual model of my home's interior. Then I install distance sensors throughout my house and add them to the virtual model in the corresponding locations. Write a program to triangulate position from sensor input, then govern behavior of integrated systems based on that location.

>>1092285

You could try RSSI with bluetooth receivers, but it's not very exact.

>>1092295
That was a good idea, a quick Google brings me to this page which makes me think it may not work well enough, unfortunately:

https://e2e.ti.com/support/wireless_connectivity/bluetooth_low_energy/f/538/t/349299

>>1092294
that is overdoing it

just use motion sensors and use it to trigger x y z

>>1092303
I guess. I was hoping to create something more overarching that could also gauge power consumption and stuff like that. But you're right.

File: Math-Nerd.jpg (615KB, 1100x825px) Image search: [Google]

615KB, 1100x825px

>>1092285
I'd like to interject for a moment to point out that the correct term for what you are referring to as triangulation is in fact trilateration.

Triangulation is where you determine position by measuring the angle between yourself and other points.

Trilateration is where you determine position by measuring the distance between yourself and other points.

>>1092305
Good point. Thank you.

File: light-switch.jpg (10KB, 618x1000px) Image search: [Google]

10KB, 618x1000px

>>1092292

>>1092297

Yeah, you really need ToF and triangulation to get a good fix.

Indoor location is still an unsolved problem. RSSI triangulation is probably the most readily available approach, but not that great in terms of performance. At the other end you can put a light beacon on yourself and do essentially motion capture (e.g. Grasp lab, Valve lighthouse). And there are funky things in between like this sonar hybrid thing out of MIT.

>>1092297
Can confirm. I've developed bluetooth low energy occupancy sensors professionally. They're good enough to tell if someone is in a room, not where someone is in the room. BLE wasn't designed with trilateration in mind. Your own small body movements will throw of RSSI by +/- 30%. A lot of people also forget that RSSI is a rough estimate of signal strength in dBm which is a logarithmic scale. Being able to tell a phone is 30cm away doesn't mean you can go out linearly and tell if it's 60cm away, 90cm away..

There is a solution of using a large number of BLE beacons. You know where each of them is and can figure out which is closest to the phone with reasonable accuracy.

one solution is to deploy kinect sensors in each room.

>>1092320

It's a solved problem, RF ToF.

It's just that the COTS stuff available to do it isn't cheap yet.

Hi, idiot here.

If all the rooms in his house have arches and door openings, could he put in coils like those at the exits of stores? two coils, one on the side of each room, so the system knows if coil 1 and 2 were triggered, or coil 2 and 1, to know where he is. Then RFID could work as far as my puny brain understands it.

I suppose it would potentially misregister which order they were triggered in, but have no idea of the rate of precision. If I had time and space to do this I'd experiment myself, probably by removing the decorative planks (sorry, ESL) on the walls of the door frame and putting the coil assembly in there.

>>1092307
will that dim leds but, and order me a case of washing-up liquid, a bottle of vodka and a fair maiden, solely based upon the angle at which my head hangs in the WC?
>inb4 its hacker-proof tho.

>>1092285
Paint registraton marks on all your walls and ceiling, make a helmet with few cameras and a 3 axis accelerotometer on it, using video from the cameras, and the acceliromiter readings determine location and orentation.

wear a toe ring with a weak rf transmitter and tile your floor with sensors

Assuming your rooms are square/rectangular, you could use 2 cameras. A camera on one wall with a wide angle lens to cover the whole room feeds video into some processing device (Raspberry Pi?). Use image recognition to track your position across one axis. The same with another camera and you could get some kind of XY position within a room.

But really if you're doing that, you could probably use 1 camera on the ceiling.

>>1092489
I want my family to be able to trigger it as well, so this idea is my favorite so far.

>>1092500
Or a cock ring, and line the walls with sensors dick high.

Try hijacking some kinect gear maybe? The hardware has range finders as well as camera, there's a software developers kit available for it too. I imagine it's already primed for detecting people, you could throw in some gestural or voice commands in there

Check out the Decawave UWB time of flight modules. Compact, 25$ a pop, 10 cm accuracy. Probably the best COTS solution out there at the moment. Atmel has a cheap chip, but it's carrier phase based which is much more sensitive to multipath and it's woefully poorly documented. Pulson/kio has something, but the kits are fucking expensive and they are way bigger than the Decawave modules.

https://sites.google.com/site/wayneholder/uwb-ranging-with-the-decawave-dwm1000

>>1092829
I'm impressed!

Atmel's method measures the phase shift between transmitter and receiver, and then measures it again using a slightly different frequency. Then by subtracting one offset from the other, can determine how many wavelengths difference there is. It is narrow band 2.4ghz though, so like you said, it sensitive to multipath. They also guard their methods in precompiled libraries, only available for the ATMEGA line (though I never asked if the source is available with a non-compete).

UWB is time of flight based. It has great accuracy due to the wide band. There is a company called timedomain which has 2cm accurate radios (which can also do UWB radar). I've played with decawave too. It really is amazing.

Here's the tradeoff though. To be accurate, an RF distance measuring method must be as close to line of sight as possible. The atmel chip is inaccurate, because it is also measuring reflections and can't tell the difference.

I got to 300 meters with a decawave with clear line of sight. I tried it in my house (which is only wood), and i couldn't get more than 15 meters if I had to go through a wall or two. If there was a metal file cabinet or something, forget it.

Also, bear in mind the human body is a good blocker of RF too. When doing my line of sight test on decawave, I spun around so the module was a few feet in front of me and my body was between the transmitter and receiver. I walked backwards towards the receiver at 300m until I got the signal back. Didn't get signal until the 30 meter mark. The antenna must be designed differently to be worn close to a human body.

>>1092909
>a non-compete
oops I meant non-disclosure. Atmel is a dead end anyway. It needs calibration per chip, nobody is using it, and who knows if they will bake it into any future radios.

>>1092909
We've had a little luck combining phase-shift distance measurement across multiple pings, always trusting the signal with the highest final strength, and limiting calculated position changes to what's physically possible... but that's a lot of data management just to get a rough location.

We had to have a high ping rate (low battery life) and multiple beacons per person (each side of the body) just to get non-useless results.

I bet the same system using TOF would improve the data considerably though. I'll have to check out the UWB modules; thanks for the link!

>>1092305
Learn something new every day. Thanks, precision of language anon.

>>1092921

Cool. Thanks for the info. I never got very far with the phase shift stuff. The network I was designing had to coexist with emergency personnel (they would also be tracked). Firefighters frequently use bluetooth on their person. Bombarding the 2.4 ghz spectrum with measurement traffic and having a transmitter right on their suit was too much of an interference risk.

Timedomain has some great papers to read, but it's organized poorly. You have to select a product and then a subset of papers are selected on the "resources" tab. This one has some cool case studies, including one with an ad-hoc localization network with a firefighter dropping "breadcrumb" radios on the ground as they move through a building.

http://www.timedomain.com/wp/wp-content/uploads/2015/11/320-0293C-Part-Four-Tracking-Architectures.pdf

Timedomain spun off a sister company http://pluslocation.com/ that takes care of setting a building up for localization. Of course, they want to own your data for you. But I found it interesting that the first step is a complete floor plan and complicated analysis to find ideal node placement to get enough coverage without dead zones due to obstacles. They also own patents on tracking just about anything inside or outside a building, so be careful if you're developing a product.

>>1092931

>They also own patents on tracking just about anything inside or outside a building, so be careful if you're developing a product.

Be careful to forget you ever heard this.

Nothing goods comes from knowing what's patented (everything).

>>1092368

you're better off with a thermal sensor in each room

a normal house's makeup will fuck with electronic signals.

A line of sight thermal sensor that detects when an object is giving off X amount of heat in a room would be better, and would likely be simpler to actually use

wouldn't want your computer or stove turning off when you took a piss, but if they generate enough heat then they'll be picked up and you'll be good

>>1092285
motion detectors in all the rooms dumbass. GPS isn't accurate to more than a few meters anyways. Only an issue if you own Hogwarts or some bullshit.

>>1094816
I own hogwarts

Use wifi and measure position based on phone

ultrasonic radar

>>1092307
kek

>>1092305
4chan doesn't deserve you. Have you considered moving to Reddit?

>>1092829
What's the range?

>>1095764

See >>1092829

So basically you would need a network of them throughout your house. 3 probably wouldn't be enough, because you wouldn't be guaranteed to see them all throughout the house.

>>1095784

Oops, I meant see >>1092909

>>1092320

RSSI is totally useless for determining distance.

Inertial tracking may eventually be the best way to do this. The space shuttle had an IMU able to go from liftoff to landing and only drift a few centimeters. There are car solutions able to work for long distances in tunnels and parking garages but these also rely on tachometer and super precise barometer information.

https://www.youtube.com/watch?v=6ijArKE8vKU

>>1092305
>I'd like to interject
Get out stallman

File: 1477107951766.png (4MB, 1599x2400px) Image search: [Google]

4MB, 1599x2400px

Triangulation may be overkill for home automation. How about sub floor pressure switches?

The concept I'm proposing would be at least two floor switches between each room on each side of the door. The order of which they are actuated indicates which room you are entering and which room you are leaving.

A higher resolution technique is used by the secret service to monitor the oval office without being in the room or using cameras

>#7, To maintain the President’s privacy, Secret Service agents aren’t stationed directly in the Oval Office. They stand outside. So if the President closes the doors, how does the Service know what POTUS is up to? They installed weight-sensitive pressure pads under the carpet — to let them know exactly where President Obama is at all times.
http://www.huffingtonpost.com/brad-meltzer/top-seven-coolest-secrets_b_7584648.html

There are high resolution pressure mapping sensors from tekscan as well as a more practical diy low resolution ones.
http://ame2.asu.edu/projects/floor/index.php

>>1092285
if you don't need precision and just want to know what room your phone is in
why not put a separate wifi ap in every room and check to which one you are connected
that should be cheap and simple with the pleasant side effect of perfect wifi coverage everywhere

File: 1481133223694.jpg (149KB, 600x1800px) Image search: [Google]

149KB, 600x1800px

>>1092779
That could theoretically work if he spends most of his time rubbing his dick on his walls.

>>1099600
How would my phone know to switch between routers? Plus having my connection change every time I switch rooms could get annoying.