HPS?

nisbus

Hi,

I'm new to this sensor scene and I am wondering if I could use the Parallax GPS module with my own transmitters (i.e. from within my home instead of from space )

I'm looking to build a Home Positioning System that has the accuracy of millimeters. Is this something that is possible?

If not the GPS module, then could I use ultrasound ?
What is the range of ultrasonic sensors/transmitters?

Any ideas would be appreciated.

Thanks,
nisbus

SRLM

I pretty sure that you can't use the GPS system, if for no other reason than it's probably illegal to generate your own signals. They are used for high risk navigation, after all. Plus, it would be incredibly difficult (near impossible?) to measure the propagation time of an RF signal in a house to within mm.

Ultrasound wouldn't work well because it would bounce off the walls quite easily, leading to bad readings.

Your best bet would be to use IR beacons and triangulate your position, rather than measure flight time.

nisbus

Thanks for your insight.

Do you know of any suitable hardware for creating IR beacons?

I would need to be able to track 10 objects in real time so I guess I would be needing 30 beacons all sending out different frequencies?

Thanks again,
nisbus

SRLM

What I would do is have each beacon in turn do the following:

1) transmit beacon number via RF
2) transmit 'start' via RF
3) swing a IR LED around 360 at constant speed.

The receivers can then calculate their heading from a single beacon, and relative position if you have two or more beacons. The system almost certainly won't give you mm of accuracy, but it should work to within a foot or so, more if you're very precise in your implementation. One side note about this implementation is that at the base level you can probably only get about 1/2Hz of position update. If you optimize the system, there isn't any real upper bound besides processor speed.

For the hardware part, any suitable IR LED and IR receiver should work. Most of the structure and all of the software will have to be custom built, but that's part of going where no man has gone before.[noparse]:)[/noparse] I'm going to try the system with 56 kHz so that it won't interfere with the lower frequency IR distance detectors on robots.

As for the number of beacons, you only need a constant number that will provide complete coverage. Each object that you want to track will obviously require it's own receiver of IR and RF signals.

kwinn

Would it not be simpler to have fixer IR beacons and have each one "blink" a different serial code or frequency and then use triangulation to determine position?

science_geek

doesnt the xbee module do signal strength, if so then get some low power ones and have maybe 10 set up around the house and a lot of redundencies in the distance measurement

nisbus

How about having a wifi receiver that measures distances from three different routers?

I live in an apartment building and my PC picks up at least 10 routers and usually those stay in the same place.
I don't know how I would plug that into my Propeller starter kit (or if it's even possible) but that would only require me to have a wifi receiver and not needing to put up beacons.

kwinn's suggestion sounds more in the line of how GPS actually works so that might be a good solution.
My only concern is the range of IR (across rooms etc).

I've never heard of the xbee module but I'll look into that.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Thanks,
nisbus

SRLM

I think the problem with measuring a signal strength is that it's such a short distance (across a room) that it doesn't degrade much. If you measure the signal strength through a wall from a source a few meters away, then it might return an identical strength to a LOS receiver 50 feet away. And your walls would have to be identical to a wireless signal: what's to say that one wall has the exact same resistance to the RF that another wall has?

IR will work across a room just fine with regular modules (your remote works from across the room, right?), and you can always get higher powered LEDs if need be. Mouser has some that are rated for communication to 60 feet or so outdoors.

Kwinn's suggestion would work, you just need one more beacon to find your position. It's a tradeoff between beacon count and beacon complexity, and up to you. Another idea you may want to consider ( and that might get you really close to your accuracy requirement) is to use a set of fixed position locations that your device periodically returns to or passes over (question: is this a mobile robot system or something else?). When the device hits that position, it knows exactly where it is at that moment in time. You can then use a accelerometer and gyroscope to calculate your position once you move away.

nisbus

Maybe an accelerometer/gyroscope combo would work best for my requirements (just skipping the triangulation completely)

What I'm trying to achieve is getting the position of my hands (or even better, each finger) in 3D space so I can send that to my 3D interface on a PC.

If we start with just 2 objects (hands) where they are stationary and press a button (space reset) it would probably be possible to measure their position according to the change in the accelerometer and gyroscope based on the original "zero" point?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Thanks,
nisbus

SRLM

You should just go with colored fingertips and use a camera to detect position. I seem to remember you saying that you're a programmer? Anyway, you could probably use something like OpenCV (take a look around in the prop forum). I'm thinking a finger tracking like this (although more bulky and not as advanced probably)...

http://www.ted.com/index.php/talks/pattie_maes_demos_the_sixth_sense.html

The problem with accelerometer and gyro combo is that you need one for each finger...

nisbus

Yeah I've seen this before but I dislike the fact that firstly the system needs a line of sight to the hands.

It would be possible to just track the hands position using an accelerometer and then track the fingers in relation to the hand.
That would also allow for a more accurate finger reading.

I've ordered one tri-axis accelerometer for testing and I'll keep you posted on my results.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Thanks,
nisbus

SRLM

You might want to try out the Wii. From a old, hazy memory I remember that you use the wand to move a cursor. You may want to try it out and see how it works, and relate it to your system. In any case, I think you'll need a gyroscope in addition to an accelerometer.

nisbus

Do you know where I could get my hands on a gyroscope that I can use with the propeller chip?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Thanks,
nisbus

SRLM

There's a prop object for interfacing with a Wii controller, so you could buy a cheap knock off of that, otherwise take a look at sparkfun. Philldapill is also working towards a IMU board, but nothing yet.

http://www.sparkfun.com/commerce/categories.php?c=85