Only in the hyperventilating startup bubble world can tracking people around a shopping mall with wifi be cooler than tracking your position anywhere on earth with satellites that were put there by huge rockets.
Which actually leads to something I've wondered about before: Why couldn't you (with appropriate FCC clearance) set up GPS transceivers around a shopping mall calibrated to rebroadcast the signal at the appropriate time offset? It would work with everyone's existing hardware and be completely transparent.
GPS is complex and I may be wrong. But for rebroadcast to work the transceiver would need to be placed exactly between the user and the satellite in space; meaning a different location every time the user moves. If not perfectly placed; the radio signal will have travelled further than expected and give the wrong position. I suppose you could attempt to locally recreate GPS signals with multiple transmitters; but this would play havoc with GPS reception outside the building. Why not just use WIFI?
Differential GPS is not about rebroadcasting the actual GPS position, but the error from a known position. This allows the device to enumerate the error caused by atmospheric effects.
The location of each GPS transmitter must be known exactly for the triangulation to work, and existing GPS receivers know nothing about the location of shopping-mall transmitters. They only know about satellites.
So that's why you couldn't do this with existing receivers, but I don't know of any reason why you couldn't make new receivers that cooperate with a new type of ground-based transmitter built into cell towers or similar. Multipath might be a problem but it seems surmountable. Do any radio experts here know why nobody has done anything like this for cell phones? (The Wi-Fi/Bluetooth positioning talked about in the article is based solely on signal strength which is far inferior to the GPS method).
I don't think the GPS approach can work with ground stations - what GPS has going for it is line-of-sight to the satellites. Multipath would be a hell of an issue in an urban environment.
Correct. The ground reference stations multiplex their data to the WAAS satellite, which then pushes the data on WAAS-specific channels (hence, why your GPS receiver specifically needs to be built to support WAAS)
The main problem is that wireless signals simply aren't reliable. You could "boost" GPS indoors, but that doesn't get around the fact that there are corners, concrete walls, mirrors, humans, and so on.
Anyway, we're splitting hairs. The point is that we don't currently have an IPS. It doesn't matter if we extend GPS to work indoors or use other signals to create an IPS -- the point is, we need to create an IPS :)
As a robotics hobbyist I've done a lot of research into the 'localization' problem (which for robot folks is localizing where you are relative to the space around you). And there are lots of ways to do it, some more interesting than others. The simplest being landmark navigation. Sebastian Thrun of the self driving car fame did some interesting graduate work in Germany where a robot uses the doorways to figure out what room it was in and then could navigate to any other room.
However what some people fail to realize is that stores really don't want this technology to exist. You see they, like Casinos and other places with 'impulse actions' want you to wander around looking for your target and being exposed to various other temptations. How much less per shopper would you imagine there would be if a shopper had a pre-planned route available for the three things they needed at a Target or a Walmart?
A really simple solution would be to program the ceiling of a place into an app. When you're looking at your phone, its front facing camera can look past you to the ceiling and figure out where you are. Roboticists have known for a while that ceilings are remarkably good navigation references, try as you might they are often quite different to the discerning camera.
How much less per shopper would you imagine there would be if a shopper had a pre-planned route available for the three things they needed at a Target or a Walmart?
How true. I was in a Target yesterday (where I almost never go) and started getting really irritated by the fact that kitchen utensils etc., were segregated by brand instead of by function. So the same type of product from two different manufacturers would be separated by 8 or 10 feet, making price/feature comparisons tedious, to say the least.
> I was in a Target yesterday (where I almost never go) and started getting really irritated by the fact that kitchen utensils etc., were segregated by brand instead of by function.
In some cases, that might be because of shipping and (re)stocking logistics. I would guess that OXO products come to the retailer in one shipment, Le Creuset products in another shipment, etc. Especially when profit margins are thin, the retailer might not be interested in paying employees to re-sort products of different brands into functional groups. That'd be even more true if stocking were done by representatives of the different manufacturers or their distributors.
You might be right, but I couldn't help noticing that if I'd purchased the more prominently merchandised products I'd have spent about $50 rather than $30. Years ago when I worked in computer retail, it was just part of the job to put all the hard disks in one spot, the network stuff in another, and so on, even though this meant extra work in splitting up the incoming shipments.
That'd be even more true if stocking were done by representatives of the different manufacturers or their distributors.
Very possibly the case, now I think about it. I know that's how 7-11 works, and I've noticed similar arrangements at Home Depot for some high-end products.
The more likely explanation is the slotting fees retailers charge to have the products in the store in the first place. It is easier to sell a bunch of space and have that space occupied by the manufacturer.
Depending on the product, the manufacturer can be responsible for filling the shelves and maintaining the display. As an example this is usually the case for Frito Lay products.
"Roboticists have known for a while that ceilings are remarkably good navigation references, try as you might they are often quite different to the discerning camera."
Would this work in an absolute sense, providing absolute location, or in a relative sense, like an optical mouse, sampling the ceiling and comparing each sample to the previous?
It ought to be good enough for absolute position. The general arrangement of lights, air vent, smoke alarm, sprinkler, motion detector, and wall proximity are more than likely unique.
Assuming this technology because popular enough to affect the stores' bottom line, how much do you think an enterprising person could charge them to install a system specifically engineered to confuse this?
I'm thinking, Portal-like attachments in the ceiling that randomly move around - probably at night, to avoid scaring shoppers.
Or maybe you could get away with a couple of computers hooked up to half a dozen projectors, projecting random imagery on the ceiling. You could even do something interesting like a rainforest canopy, clouds, etc.
I really love this visual. Black inverted beetle hump devices, masquerading as security cameras, moving from place to place on the roof at night when the lights are out to confuse nav algorithms that defeat 'enchhanced discovery' layout.
If you want to make it unique, it syould be simple to cover it wit QR codes. If you know the location, size and orientation of each (which can be encoded in the QR itself), you should be able to calculate the position of the cameea pretty accurately by meassuring the perspective distortion.
If you check out SLAM positioning, you can do some really amazing stuff even if all you know is how far away from the wall you are. Given a decent floor plan, you simply take frequent distance measurements as you wander around, using dead reckoning (the accelerometer will help) to keep track of your relative positions. You rapidly build up a local map of the walls near you, which eventually will map uniquely to one location on the floor plan. Our robot only had to wander for a few seconds to figure out where it was, and that was just using a few ultrasonic rangefinders.
I haven't heard of such a thing, but I imagine you could do something similarly cool with the camera of your phone. Or, even the accelerometer and compass, combined with a knowledge of the locations of doors and the available walking paths, may be able to give a reasonably good approximation of your current location--if you walked north for 30 meters, then turned left and walked another 20 meters, then north for an additional 5, in a typical mall that's going to limit the number of points you could be occupying.
Why not? I guess the features would have to be large/visible enough to be picked up by the camera. For initialization some kind of qr-codes could be placed wihtin the relative positioning patterns...
The article claims that " GPS and GLONASS can measure altitude, but generally the data is inaccurate and too low-resolution (on the order of 10-25 meters" - which I'm fairly sure is totally un-true.
I use GPS to fly UAVs and I've seen errors in GPS altitude up to 50ft compared to barometric altitude. Generally it's better than that, though. It's accurate enough for us to fly autonomous helicopters with the barometric sensor disabled. I'd say that disproves the inaccurate and low-resolution claims ;-)
No, it's true. The possible geometry of visible GPS satellites at any given moment (namely that they're all above you and none are below) means that the vertical error is pretty bad in comparison to the horizontal error.
It has much more to do with how GPS receivers calculate a horizontal position when they have limited satellites in view.
I don't understand the math but essentially you are trying to solve four unknowns. Latitude, longitude, elevation, and the time offset between the satellite and device clocks. If only three satellites are in view the only way to solve latitude and longitude is to assume a value for elevation based on previous results. Without a forth satellite the GPS simply doesn't know what the altitude is. But that estimated value is sufficient to give a good estimate of horizontal position.
It's true that this makes the VDOP generally higher than the PDOP, which leads to lower precision in the vertical direction. The 10-25m figure is a complete nonsense, though. It's more like 3-5m.
WAAS uses a network of ground-based reference stations, in North America and Hawaii, to measure small variations in the GPS satellites' signals in the western hemisphere. Measurements from the reference stations are routed to master stations, which queue the received Deviation Correction (DC) and send the correction messages to geostationary WAAS satellites in a timely manner (every 5 seconds or better). Those satellites broadcast the correction messages back to Earth, where WAAS-enabled GPS receivers use the corrections while computing their positions to improve [3D] accuracy.
And if you need even higher resolution (centimeter resolution), use use NOAA CORS sites (either with post-processing, or real time with their experimental datafeed)
I'm sure most of the responses to your comment have technical reasons for their disagreement, but I agree with you for anecdotal reasons.
I have a Garmin GPS I wear when I run, it connects to 3 (at least?) satellites and I've never seen it be off by more than a few feet in altitude. I'm sure there are some random points in the graph with anomalous readings but the overall "smoothed" graph seems very accurate.
Granted, there are times it loses its signal altogether, but that is not a resolution issue.
I got a certificate in GIS back in 2002. Dusting off my brain: All GPS readings are only accurate to within about 10 meters if they are quick and dirty. To get accurate readings, you need multiple satellites in view for up to several hours.
I am assuming that has not really changed. Anyone with brand spanking new credentials or more current industry experience feel free to correct/update. Thanks.
If you're in the continental United States (or Canada), and have a GPS receiver that can handle the WAAS adjustment signal (http://en.wikipedia.org/wiki/Wide_Area_Augmentation_System), you can get a much more accurate signal as quickly as a regular GPS signal.
This would be nice if it let me find the spaghetti noodles more quickly in an unfamiliar grocery store. But somehow I doubt the store is going to install a bunch of tech to help me shop more quickly.
If they wanted to do that, they could just all put stuff in the same place.
Check out what the MONA has done in conjunction with Art Processors http://mona.net.au/theo/ to completely get rid of tombstones inside a museum. Harder to do here at the Met, but there are most definitely applications that aren't about spending more money.
Commerce isn't a bad thing, but exciting new technology to try to separate me from my money faster may be a cool thing for Google, but not for me--the consumer. Maybe I'm weird, but I already know what I want to buy, I don't need technology to try (and usually fail) to generate some new consumption desire.
Be honest though - it's not like companies like to develop technology that has no commercial value. Whether or not it's used in that way is a different question, but if it's a big company developing something, you bet it went through twenty levels of market research to make sure there is a buck to make at the other end.
It's still your choice what you do with it once the technology gets to you. I find it far more fruitful to look at what people do with the technology and how they end up screwing themselves with it. And by fruitful I mean frustrating.
I cannot wait for this to get some adoption! I love augmented reality, but the current state-of-the-art is not exactly suited for AR applications. If this chip can deliver on its promises, I think we'll see an explosion in AR as it finally starts to look real. The main problem with AR on mobile devices as I see it is a visual one - very intense jitter due to inaccurate GPS and gyroscope data. If IPS can pinpoint AR objects not based on visual markers or inaccurate GPS, then I think we are quickly headed into Dennou Coil territory. (http://en.wikipedia.org/wiki/Denn%C5%8D_Coil)
I too cannot wait for true AR. I have been dreaming about making AR apps for the iPhone since it was released, but have never found satisfactory solutions to the positioning issues. Can't wait to play a real world RPG, with GPS used for the world map, and IPS for dungeons with AR objects accurately placed.
There isn't a standard yet; wifi locationing currently requires a site survey to gather location data. It's potentially possible to crowdsource this data, but currently indoor location tracking has nothing like a GPS standard – each site is different.
Think wi-fi is cool? Sky-fi will blow your mind! It's just like wi-fi, only it uses satellites instead of terrestrial nodes for communication.
That is to say, my main problem with IPS is that it sounds like GPS, but isn't at all the same. One is a massive global satellite network, and the other is glorified RFID with a hefty glob of marketing on top. Call it something else, at least, please?
Just because the article shows examples of how that tech is applied to consumerism doesn't mean thats all IPS is limited to. I'm sure you can think of many applications that are not solely to make more money.
A cool system would be a tool to find the location of a point relative to the sun. So I can whip out my smartphone while in the asteroid belt and have an intuitive metric system to identify any point in our solar system.
You'd need at least two more points of reference, I think, in order for this to be effective. By itself, a picture of the sun (and I'm assuming it's an incredibly clear picture taken with a magical instrument) would only tell you how far away from the sun you were.
With an additional known point (and an accurate date and time), you'd know where you were within a circle around the line between those two points. That might actually be good enough for a rough estimate, if you can assume that you're in the plane of the solar system.
