GPS granularity / accuracy in latest smartphones

[gagan_goku]

Hi guys

I am thinking of building an application to detect which user i'm pointing towards using my smartphone. Imagine that multiple people are present in a big room, all of them with their GPS enabled, and i point my smartphone towards one of them.

I wanted to know what is the accuracy of the current GPS hardware and software in our smartphones like iPhone / android? Is it possible to draw a line in space using the location pointed to by GPS and the direction it shows accurately to within say 1 meter? I know that the accuracy degrades indoors, but how much error do you think there would be in a bar for example?

 

[HavocXphere]

You've got it all back to front. GPS doesn't work like that.

And no, what you're thinking about won't work. For starters GPS indoors is kinda sketchy...

 

[gagan_goku]

You've got it all back to front. GPS doesn't work like that.

And no, what you're thinking about won't work. For starters GPS indoors is kinda sketchy...

which part, might i ask? GPS receivers in our smartphones give us the latitude longitude within a certain degree of accuracy, which is what im interested in. I can figure out the orientation of the phone using Compass.
Now the only question is how accurate are these measurements.

 

[HavocXphere]

which part, might i ask?

The part where latitude & longitude is not equal to second by second direction.

GPS is stylish, but if you're looking at meter to meter accuracy directional measurements then GPS isn't going to cut it. At least not the ones in phone GPSs. The architects use GPSs that can pull that, but thats no use to your situation.

You might be able to combine gyroscopes & GPS, but I'm telling you right now its a long shot, esp if you don't standardize on gear manuf e.g. Apple.

I still get the impression that you don't understand what GPS is & what it does. We're looking at a couple of meters inaccuracy at least, so little chance of direction being accurate within a room. In a car moving many meters in a couple of sec - sure. In a room - not so much.

 

[gagan_goku]

I'm not looking to do anything with directions or navigation. My use case is the following: a user is sitting in a room. He wants to know more about the user sitting across 2 tables across.

I still get the impression that you don't understand what GPS is & what it does. We're looking at a couple of meters inaccuracy at least, so little chance of direction being accurate within a room. In a car moving many meters in a couple of sec - sure. In a room - not so much.

Does the measurement stabilize if the device is stationary? Or is there some other technique i can use (without using additional hardware) to get a more accurate idea of where i am in a room ?

 

[HavocXphere]

I'm not looking to do anything with directions or navigation. My use case is the following: a user is sitting in a room. He wants to know more about the user sitting across 2 tables across.

2 tables across...nice. *Which* 2 tables across would that be? The restaurant has 20 tables. Hence directional.

Does the measurement stabilize if the device is stationary? Or is there some other technique i can use (without using additional hardware) to get a more accurate idea of where i am in a room ?

No. You need different (expensive) gear to gtt more accuracy.

 

[gagan_goku]

Yes, i was hoping to decouple the gyroscope sensor from the GPS. Anyways, was worth a try, would have made a nice application had it been reliable
Thanks for your inputs HavocXphere.

 

[jem]

you're more likely to succeed with some sort of face recognition rather than GPS/Gyro.

 

[gagan_goku]

except thats a much harder problem, plus taking photos might not sit well with some people.

 

[cpu.]

Drive with a GPS on a highway and take an exit running alongside and then away from the road you came from. Check how the GPS re-act. It still follows the original road, then as you move away adjust after a few seconds.

I think it's illegal to have a GPS with pinpoint accuracy (without a license).

 

[ozziej]

GPS will not work. As many have said their accuracy is too low, plus, you don't know which person you are pointing at. Are you pointing at joe that is two meters in front of you or sue that is another meter behind him?
Have you thought of doing something rather simpler? You could perhaps have an application on each persons phone, that will then broadcast their details, you could then use that phone's gyro and compass along with you gyro and compass to determine when the two phones are pointing at each other. You could then use something like Bluetooth to transmit data between the two...

 

[shogun]

A bit late to the party, but everyone else above is correct. No hope in hell of doing this with GPS. Not indoors, not even outdoors unless the distances are much greater. GPS accuracy can be anywhere from 1-2m to 50m, depending on what is getting in the way of the signals. Indoors, the only signal you are getting is from whatever radio waves happen to bounce through the window (they don't travel through most roof materials).

This means that even if you even pick up some signal indoors, you will experience severe multi path failure (GPS uses the time between signals in the radio waves of multiple satellites to determine location).

Even if you had to place your tables outdoors in a perfect reception area, you will still be influenced by GPS drift (the position will start to drift around a point). GPS positions will always drift when stationary, and this drift is usually a few meters around your current location. Bear in mind that all the GPS's at your tables would drift as well, and your error increases greatly.

Perhaps in Europe or the US where you can get an SBAS like WAAS or EGNOS, you can get better accuracy (outdoors), but that would only help with a decent GPS antenna in any case. The problem with cellphones is that they have a tiny patch antenna, which means that they are doomed to failure when it comes to high accuracy in any case. When it comes to any radio, the bigger the antenna (of the same design), the better.

The system you are proposing sounds awesome, but you'd need another technique (and likely hardware). The one thing that comes to mind is radio triangulation. I saw a system for tracking cats. Each cat has a tiny radio transmitter, and then you get a hand held tracking unit where the signal gets stronger as you point it towards the cat. How you'd be able to use the hardware in a phone, or determine distance accurately is another challenge altogether.

 

[daffy]

Everyone's forgetting the obvious detail here.
How does my Phone know the GPS location of the guy sitting at the other table?

 

[HavocXphere]

I think it's illegal to have a GPS with pinpoint accuracy (without a license).

Nope. No SA laws regulate GPS in any manner afaik. However:

You get military access and civilian access. Right now both provide the same accuracy. Civilian has restrictions though designed to prevent use in a DIY cruise missile. i.e. If you go to fast or too high it stops working.

The mechanism that added noise to the civilian channel was disabled a couple of years ago by US prez executive order. If war time comes they'll switch it back on again and then all our GPSs will be wildly inaccurate again.

Pinpoint accuracy uses a slightly different underlying tech and doesn't cope well with movement. i.e. Engineers put it somewhere for measurements. Then move it and take 2nd measurement etc.