If you're navigating on a boat, every boat has it's own deviation chart. When you then use the compass you factor in variation (what the chart says the difference is between mag north and true north) and the personalized deviation chart for that boat.
To set this up, some guy comes down to the dock, takes your boat out, does a few 360's with known landmarks in site and compiles a table of how your boat's compass varies from standard magnetic north. Big things like the engine, generator, keel, etc will influence what your deviation chart looks like. Usually it's less than 4 degrees at each point of the compass IIRC.
So every place in the world has metal objects that create that place's own distinct deviation from where magnetic north should be. So perhaps that's what these guys are using. A known table of variation and then looking at the deviation.
What I wonder about is what happens when someone turns on a 12 amp vacuum cleaner, a monitor nearby goes into powersaving mode, someone moves their laptop or someone with their own cellphone walks past. These things emit magnetic fields, so I wonder how they've solved the problem of the local deviation changing constantly.
As famousactress may have been alluding to, this is equivalent to the well understood problem of robot localization in dynamic environments. If my robot has a map of the building and it thinks it knows where it is, will it get confused if someone walks in front of it? Not if I reserve some probability mass for the possibility of seeing an obstacle at any distance (in the range of the sensor).
> What I wonder about is what happens when someone turns on a 12 amp vacuum cleaner, a monitor nearby goes into powersaving mode, someone moves their laptop or someone with their own cellphone walks past. These things emit magnetic fields, so I wonder how they've solved the problem of the local deviation changing constantly.
Are any of those magnetic disturbances even within an order of magnitude of a smartphone's compass threshold? I'm surprised that the disturbance from a skyscraper's skeleton would be detectable, and a computer monitor seems like even more of a stretch.
It's clear some disturbances must be measurable because as I walk around the room, the magnetic field measured by my smartphone constantly changes and since the magnetic field of the Earth is practically uniform over small distances, this must mean the disturbances come from my surroundings.
However, when toggling some apparatuses in my room, nothing happened. My laptop doesn't have a hard drive though, but even only centimetres from it, no change was measurable. Neither did anything happen near my lamps or the AC/DC transformator. The only place I measured a difference (<10%) was on the stove (why yes I'll put my smartphone worth hundreds of euros on a hot stove). However, this faded into the background at a distance of about 10 cm.
It's possible that constant change would be dealt with by probability and consensus. Compared to your best historical guesses at this location, how likely is it that the reading you're getting is accurate? You can establish high or low confidence and disregard sudden changes. Also, the more nodes (people, phones) you have the more you can build consensus between them, use the higher confidence predictions as waypoints for 'fixing' lower confidence ones.
I have a small site dedicated to providing information about gps/cell phone tracking. As part of that, I've been doing a bit of reading about cell phone tracking capabilities.
It amazes me that we are turning our phones into better tracking systems than we've ever used on wild animals. Whether it's sound fingerprinting, gps, WiFi tracking, or now magnetic disturbance tracking, your phone's location services is getting so it can tell within a few feet where you are at all times. (By 2016 FCC regulations make it a requirement that all phones locate themselves within 100m or so)
This has amazing potential for startups. If your phone knows you are at the gym, it can start your workout routine. If you are at the library, it can decrease the ringer volume. Now with inter-building location ramping up, the opportunities will only increase.
People wearing little electronic devices that allow law enforcement to determine their position and travel patterns over any period of time just by filling out a form and without a warrant. If you had told me 20 years ago this is where we'd be, I would have called you a paranoid nut-job.
One of the things I was curious about is whether the phone can be enabled remotely by the telco. I found conflicting information on the web. I suspect it can, but this is not something the carriers want to talk about. Just a suspicion, though. I imagine what happens 99% of the time is that the user doesn't want anything to do with tech like gps tracking, but then they find an app that does something cool, and it things like gps to be turned on in order for it to work. So in a way, just like Facebook is destroying privacy by "helping" folks share, many apps are destroying the anonymity of location and travel by "helping" folks with cool apps. That kinda sucks. Wish the situation were different.
Note that location services comprises many different technologies, not all of which can be turned off. Many of them are required for the phone to operate. The upcoming FCC regs, for instance, require 100m self-locating ability to always be on. I guess for things like 911 service?
Speaking of paranoia, there's also rumors that the FBI/black helicopter/MIB bunch can actually power-up your phone remotely, especially with some models. This sounds completely out-of-left-field to me, but who knows? Court docs show they can use your phone as a listening device even when you're not calling somebody, so I wouldn't put other things along these lines past them. There's probably a good reason Osama Bin Laden refused to have anybody associated with him possess a cell phone, whether it was used, had a battery in it, or not. Seems like I read something somewhere once about illuminating electronics gear with microwaves, then reading the signature of the radiation emitted. But it could have been in a pulp sci-fi novel. As I said, it's difficult to tell where reality ends and paranoia begins with this because reality is quickly catching up to the paranoia of just a few years ago. Who would have imagined sub-meter resolution on where you are? That's almost accurate enough to tell if you're wearing the phone in your jacket pocket or on your belt. Crazy stuff.
I would think the capability of having an Off phone receive a radio signal to turn itself On would require pretty deliberate hardware design.
I do remember reading about locally fingerprinting an Off phone based on its passive radio response, and this seems feasible.
When the phone is On, total location privacy is out the window, as Ma Bell always knows what towers you're near and can triangulate. Everything else you've said can be (and probably is) implemented in invisible layers of software.
End-to-end privacy would be a nice step forward, but assuring this requires an auditable interface between the radio transceiver and the computer/sensors. Any fix for the location problem involves decoupling identity/billing from the physical infrastructure.
>I would think the capability of having an Off phone receive a radio signal to turn itself On would require pretty deliberate hardware design.
Wake on LAN (WoL) is a pretty common feature for network chipsets. I don't know anything about mobile networks (so take this as the semi-educated speculation that it is), but it wouldn't be a stretch to imagine WoL over 3G. It wouldn't have to be a feature advertised to end users of the SoC.
What it comes down to is "what does 'off' mean on a cell phone?"
The SIM and baseband have absolute control on the rest of the phone, which doesn't make sense except for enabling this kind of surveillance capability. False/partial shutdowns or spontaneous wakeups of the radio portion would probably be noticed by this point, so it doesn't make sense that either would be enabled except in a targeted way.
From 5.4.3, it actually looks like this kind of magnetic positioning system works better in an indoor environment with lots of magnetic perturbations. This method measures where the person has moved, not where they currently are against a fixed constellation (like GPS). I'm guessing you would need to activate the data collection at a known waypoint (e.g. the front door of a business) and then it could track your movements around a magnetically perturbed space.
Those are stationary perturbations they are talking about, the ones they use to map the region. Your parent refers to moving perturbations, such as vehicles or electric motors.
An underground mine is really really really different to a subway network if you are only looking at it magnetically.
It apparently needs the user to move to get a position lock.
This will not work well with the typical use case: user is in a shopping mall, needs to go to the bathroom, wonders where the closest one is and where he is, stops walking, opens his phone and starts the app.
Expected outcome: the app says "toilets are 50 meters to the right".
If the app now tells the user "I have no idea where you are. Please walk in a random direction in a straight line for 20 meters and I will try to find out, then tell you where you are and where the closest toilets are, which means that you might have walked in the wrong direction to begin with", the user is not really happy.
If this causes a negligible drain on the battery, then the device could passively record the magnetic field at all times into a circular buffer. When they ask where the toilets are, the device can check the recent data to immediately figure out where they are.
I really like this new approach to the positioning problem, especially since it doesn't require setting up wifi or shooting satellites into orbit.
Does anyone know why this couldn't be used for outside positioning, much like GPS? Is the information not detailed enough, or do we lack good enough sensors to detect the fine-grained differences especially given all the electromagnetic noise that modern cities for example create.
Edit: I just realized it's probably also because somebody would need to create a world-magnetic-field map, much like the street view cars driving all over the world.
Which we have, not only is there a coarse scale global magnetic field model (two versions, in fact) there are detailed high resolution maps for large areas of the world (eg: all of Mali, all of Australia, most of Canada, much of the central 'stans) where mineral exploration has taken place.
The positioning is complicated by the drift over time (there's a "new" global model every five years and for specific times & locations interpolation is performed) and fluctuates daily (as the Earth turns there's a diurnal pulsing in the induced magnetic field).
This is where "a bit of signal processing" comes into play.
The good news is the maths, the field testing, and the application of it is all a good 50 years old.
There are solutions that do not need any pre-established magnetic field maps, because they simply derive velocity from local magnetic field disturbances and use magnometer measurements as an additional input for odometry calculations, using data fusion algorithms (e.g Kalman filtering). This allows to basically eliminate the drift induced by two successive integrations with low-cost IMUs.
And with enough sensors (4 magnetometers) you can reconstruct 3D trajectory.
Chapter 5 is very instructive, in particular the section Measuring magnetic fields gradients to derive velocity.
Excerpts:
If the body moves, then the sensed magnetic field must change according to Maxwell’s equations. If the magnetic measurements do not change significantly, then the solid body is not moving. This permits us to rule out velocity drifts in our estimation. Ultimately, this improves the position information obtained by integrating the velocity estimate.
Note that they don't use the raw magnetic field, but its gradients, i.e a physical quantity that is intrinsequely local. Filtering out the earth's magnetic fields becomes quite easy (I suppose..) since these gradients are fast-varying quantities in comparison. Kind of like when you use an accelerometer: gravity is low-frequency, motions are high-frequency.
Very interesting link... From 5.4.3, it actually looks like their magnetic positioning system works better in an indoor environment with lots of magnetic perturbations.
I would not have expected that result, but thinking about it more it does make sense. They are estimating "where have I moved" using the local magnetic field, which is very different than how GPS works "where am I in relation to a fixed constellation".
Whoa :) This is very clever. I've been working on indoor positioning quite a long time already and might give this a try.
No WiFi -> no extra WiFi hardware costs -> widespread use
This technology might disrupt indoor advertisement.
I guess this would be a relatively relevant place to ask, does anyone know of an affordable technology that can be used for precision (inches) tracking of 8-36 items indoor or out? The caveat would be affordable. The use case would is BMX Racing. Scoring is still done by hand! I've looked into RFID tagging but the docs and manufacturer's specs are all over the place. Any ideas?
Note: I will keep my comments to passive RFID (non-battery powered tags) and leave aside active RFID commonly used for toll roads, shipping containers, etc.
RFID is actually a family of technologies, typically defined by ISO standards, and operating over several distinct frequencies: 134kHz (LF), 13.56MHz (HF), and 900MHz (UHF). LF and HF use inductive coupling between the tag (transponder) and reader (transceiver) while UHF uses capacitive coupling. Theses technologies do not allow one to know a specific distance, but rather when a tag is in the field or not. So, when Wal-Mart uses UHF tags on their pallets, they don't know where the pallet is exactly in their warehouse, they simply know when the pallet has passed through their loading dock. I've seen some triathlon timing systems use UHF RFID, but they still use cameras in case of close finishes.
Thanks! I've seen some timing systems that are used in triathlons but the main problem was making a field that was 25-30 feet width but a tight spread (1 foot). Do you think active would fair better in that respect? Last time I checked it seemed so but it also jacked up the price a bit. I know that for Olympic BMX they are using some type of active transponders but I can't think that any commercially available solution would be accurate enough. I assume they have to be doing camera checking too but for .000 second split times even a foot spread seems like a wide margin of error.
Given the granularity in timing (i.e. 0.000 sec) you are looking for, I don't think active would help; and, the tags are certainly more expensive, though reusable.
I don't know about the affordable part, but there are a number of timing companies that offer very accurate timing within thousandths of a second.
The two companies that are the most prevalent are MyLaps [1] and Westhold [2], although there are some others out there as well. Both of these are active systems that require that each competitor have a transponder on their vehicle.
Does it need to be real time? you could mount fiducial marks on posters and capture them using head cams. Use computer vision algorithms to guess location.
I just don't see indoor navigation as a huge problem to be solved. And this seems a phenomenally complicated solution to a small problem.
GPS works because outside is an enormous, unfamiliar place with tens of billions of destinations. Inside is small, relatively familiar, and even a large building has, at most, hundreds of destinations.
Just watching the demo, it's super cool and super intelligent. But it doesn't look like it can do direction - it can only tell direction after you've walked a certain distance. How will it cope with large magnetic things that move - eg Forklift trucks, vehicles, trolleys, wire cages etc.
I don't buy it. Ikea is the only potential place I could see this being of use, and there's no way anyone's going to bake in an extra sensor to a smartphone on the offchance. Look how slow adoption NFC has seen and that's a MASSIVE problem with multiple applications and a clear financial incentive.
I suppose you don't live in a dense city with a lot of indoors and underground infrastructure. In some places you easily spend half a mile indoors between underground train platforms or to the exit on the surface, or in walkways inside and in between buildings where there's no reliable signal. Other times it's simply because part of the commute goes through tunnels or the entire trip is underground. This really breaks a lot of useful tools, for example location based alerts or simply using navigation to get to the correct exit of a station (office, mall, etc.) which if you get it wrong can mean an extra mile of walking.
Yep, location-based alerts such as when you have reminders using 'geofencing' (e.g. remind me to xyz when I leave work) -- they could fire as you leave your actual office, instead of after you drive away (which is often not as helpful).
We build indoor navigation systems for hospitals in Canada and it is indeed a huge problem. The larger institutions are like a maze with wings and additions added over the course of decades. Making a route to your appointment though 3 buildings, up 2 elevators, and over a wing is easy. Helping you actually stay on the route as you follow it is not. If this new tech is cheaper to deploy than dozens, if not hundreds, of wifi beacons then it could be a real win.
Sorry, your assumptions are way off. Indoor mapping and specifically location aware indoor mapping is huge.
We're trying to solve exactly this problem in London, why you may ask? When you have millions of commuters using your network each day (Underground) and many of your stations are at 100% capacity then your only option is to manage the flow of passengers more efficiently.
You've also missed a key problem, accessibility. The sheer volume of tourists, elderly or parents with prams getting onto a carriage thats inefficient to their travel results in exacerbating delays on platforms, this example(and other technologies) is a possible way to reduce those problems.
Working for a medium size manufacturing plant that wants to get a handle on movement of goods and people, I can tell you that this technology would be a game changer. We can't afford to use RFID and 1m accuracy of Wifi positioning is not worth the hassle. This site says they can get .2m accuracy, no extra infrastructure required, we will be watching this technology very closely. It is a lot easier to stick an iPod touch in someone's pocket than install RFID infrastructure.
I am talking about movement of goods within a single warehouse. We could stick an iPod touch on the bottom of a pallet/box for a couple days. If the tech takes off maybe someone will invent a ruggedized tiny piece of hardware that just has the compass and wifi data transmission built in.
What extra sensor? The article talks about them using the compasses that smartphones already have.
And you're wrong, it's a big market. Every large store in the world wants customers to easily find things. Big-box retail has to be over $1 trillion in annual revenue; Walmart alone does upwards of $400 billion a year. Something that slightly increases purchases or customer satisfaction can have a giant impact.
Those places employ millions of people; just letting their own employees find things easily could save a lot of labor time. Ditto for letting employees do something more productive than helping people find things that their phones could lead them to.
I'm wrong about the sensor, fair point. Though low-level access to its data will presumably need to be available through the device API.
But anyway... Stores actually put a lot of effort into spreading the essentials around, and making you walk as far as possible to find things. It's in their interests that you are lost and browsing all the other stuff that you could be buying, instead of just grabbing that milk and going.
It's in their interest that you walk past things you might need to trigger spontaneous purchases, I agree. But it's not so clear to me that they really want you wandering and frustrated. Note that most stores label aisles clearly, keep things in the same place for long periods, and often provide ways to find things yourself (staff, phones, mobile apps, paper maps).
I looked into indoor navigation-related stuff when I was planning my senior project in college. There seems to be a problem on some university campuses where freshmen, visitors, etc. get lost easily because buildings get merged with other buildings and the university can't be bothered to renumber rooms.
I don't think indoor navigation is as big as NFC, but I can see people making at least a little bit of money at colleges (and, as someone else pointed out below, mass transit situations)
Indoor navigation is a huge problem for the blind and visually impaired as well. Currently, I'm working in research on an RFID based location sensing system. It works, but requires infrastructure to be installed, so any system using just a magnetometer/inertial system would be very useful.
I'd like to use some indoor navigation at large airports and subway. And it's a killer feature for exploring underground mines and caves, urban explorers will love it :)
Fortunately there are others with more imagination.
If all you can think of is Ikea, maybe hold off commenting until (you've actually read the article) others have opened your mind to more possibilities.
How does this work? I understand if you know the true north and the magnetic north, you can get the difference, but I assume a compas only can measure one of them (the magnetic one). How do they determine the 'fault' in the signal with just 1 signal?
Every phone using the app can contribute to your map data. If you combine magnetic data plus wifi data plus user behavior data, I expect you'd be able to continue to provide good 3D data to users.
I'd bet the hard part is getting building owners to update things regularly. Who's going to remember to change the map when the grocery store moves the cereal from aisle 2 to aisle 7?
You raise another point I haven't thought of: 3D. People are different heights... surely that changes the fields too. Interesting problems to be overcome.
It must require you to be moving in a straight line; otherwise you have just one parameter, the angle of the magnetic field, and there is no way for that one parameter to specify your location in 2 dimensional space.
I would guess they would have to use the accelerometers and gyros on the phone to get a plane curve, along which they would then have the gradient and could match against a stored database.
I always thought a reasonable solution to this problem was triangulating via wifi and bluetooth signal strength from access points of known location. I tried it as a weekend project a few years ago but random fading was a killer and I never got it to work that well. Seems like it should be possible though. Are there good reasons not to do it this way?
Not everywhere has wifi and bluetooth (yes, places exist outside of the wired world) and from the article it seems that an application area of interest is underground mining where transmitting radio through tunnels is already problematic.
Seems similar to the technology demonstrated in this article [1]. However, it does seem that IndoorAtlas Ltd had a better opportunity recognition radar.
With reservations --simply because I don't know their details-- I'll say that this is nothing new at all and it could be the makings of another case of patent abuse.
Local changes to the magnetic field have been used for quite some time for purposes such as oil exploration. I remember talking to an old geologist/oil man about twenty years ago as he went about locating candidate areas to explore based on these fluctuations. He explained that they'd use magnetometers to map out underground "cracks", deposits and other features. To go from that from generating local location data by overlaying a map of some sort (whether it is a building or something else) is nothing less than trivial.
If you have any experience sailing you may have also witnessed the effect of large sunken metallic objects on the compass. Again, you could "navigate" by these effects "Hey we must be passing over the SS-Sunken Ship".
Given that mobile phones have wires and electricity then they are creating small magnetic waves which will impact others.
That all said this technolody is very suitable for underwater GPS were a GPS signal does not penetrate the water due to weakness and how bad radio travels thru water in general.
So given that I would have to questions if the military don't already use something very similiar in submarines. Lets say I'd be very supprised if this was not being already utilised in some form or another in that feild.
To set this up, some guy comes down to the dock, takes your boat out, does a few 360's with known landmarks in site and compiles a table of how your boat's compass varies from standard magnetic north. Big things like the engine, generator, keel, etc will influence what your deviation chart looks like. Usually it's less than 4 degrees at each point of the compass IIRC.
So every place in the world has metal objects that create that place's own distinct deviation from where magnetic north should be. So perhaps that's what these guys are using. A known table of variation and then looking at the deviation.
What I wonder about is what happens when someone turns on a 12 amp vacuum cleaner, a monitor nearby goes into powersaving mode, someone moves their laptop or someone with their own cellphone walks past. These things emit magnetic fields, so I wonder how they've solved the problem of the local deviation changing constantly.